Software Built for AI Agents contains 61 companies across
8 categories.
This landscape covers startups building software whose primary user is an AI agent rather than a human. These products give agents the infrastructure they need to act independently across digital and real-world environments.
Identity, Credentials & Permissions
Software that gives agents a persistent identity and controlled access to external systems while keeping credentials and permissions secure.
Multifactor
Website: multifactor.com
Multifactor builds zero-trust authentication, authorization, and auditing software that lets humans and AI agents securely share and use online accounts without exposing underlying credentials.
Category: Identity, Credentials & Permissions.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Identity, Credentials & Permissions because it gives agents a persistent, controlled identity and revocable access to external systems while keeping credentials hidden and permissions scoped.
Funding rounds
-
seed
announced 2025-12-01
: 15000000 USD
; investors: Nexus Venture Partners, Y Combinator, Taurus Ventures, Honeystone Ventures, Flex Capital, Pioneer Fund, Ritual Capital, Liquid2 Ventures, Mohan & Padma Warrior, Gokul Rajaram, Mathilde Collin
; source: PR Newswire
; https://www.prnewswire.com/news-releases/yc-f25-startup-multifactor-raises-15m-seed-round-to-make-online-accounts-safe-for-ai-agents-302633496.html
Company timeline
-
2026-07-23:
Current website packaged Multifactor as a secure account manager
(gtm)
. The site now sells Multifactor as a unified secure platform with instant access control, fine-grained permissions, detailed event history, and custom pricing for large customers. ([multifactor.com](https://multifactor.com/))
Source: Multifactor Homepage and Portal page.
https://multifactor.com/
-
2026-07-13:
AI reconnaissance login automation introduced
(gtm)
. Multifactor described Checkpoint using AI reconnaissance to build deterministic login flows and replay them inside a trusted execution environment. ([multifactor.com](https://multifactor.com/blog))
Source: Multifactor Blog.
https://multifactor.com/blog/teaching-software-to-log-into-any-website
-
2026-07-13:
Checkpoint advanced deterministic AI login flows
(product)
. Multifactor described Checkpoint as using AI reconnaissance to build deterministic login flows and replay them inside a trusted execution environment, without exposing credentials to AI. ([multifactor.com](https://multifactor.com/docs?utm_source=openai))
Source: Multifactor Blog.
https://multifactor.com/blog/teaching-software-to-log-into-any-website
-
2026-06-01:
Keypo acquisition expanded the platform
(gtm)
. Multifactor acquired Keypo, a programmable encryption company, and framed the deal as strengthening its zero-trust security stack for AI agents. ([multifactor.com](https://multifactor.com/blog/keypo-multifactor-founders-reflection))
Source: Multifactor Blog.
https://multifactor.com/blog/keypo-multifactor-founders-reflection
-
2026-06-01:
Keypo acquisition added programmable encryption capability
(product)
. Multifactor acquired Keypo, a programmable encryption company, and said Keypo’s founder would join as Founding Engineer. ([multifactor.com](https://multifactor.com/blog))
Source: Multifactor Blog.
https://multifactor.com/blog/introducing-checkpoint
-
2026-06-01:
Dave Blumenfeld joins Multifactor as Founding Engineer
(hr)
. Multifactor announced that it acquired Keypo and that Keypo founder Dave Blumenfeld is staying on as Multifactor's Founding Engineer.. Team size observed: 11-50 employees employees.
Source: Multifactor Blog.
https://multifactor.com/blog/multifactor-acquires-keypo-to-accelerate-zero-trust-security-for-ai-agents
Arcade.dev
Website: arcade.dev
Arcade.dev is an enterprise-ready actions runtime for AI agents that handles agent authorization, OAuth/token management, and governed tool execution so agents can safely act across business systems.
Category: Identity, Credentials & Permissions.
Funding stage: Series A.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits the identity/credentials/permissions category because its core product manages persistent agent authorization, user-scoped permissions, OAuth credentials, and controlled tool access for AI agents.
Funding rounds
-
Series A
announced 2026-06-01
: 60000000 USD
; investors: SYN Ventures, Morgan Stanley, Wipro
; source: Arcade.dev
; https://www.arcade.dev/blog/arcade-series-a/
-
Seed
announced 2025-03-01
: 12000000 USD
; investors: Laude Ventures, Flybridge Ventures, Hanabi Capital, Neotribe, Andy Rachleff
; source: Arcade.dev
; https://www.arcade.dev/blog/arcade-dev-raises-12m-to-solve-the-biggest-security-challenge-in-ai-agents/
Company timeline
-
2026-07-23:
Hiring cluster expanded across engineering, product, and GTM
(hr)
. Arcade.dev’s LinkedIn jobs page currently shows open roles including Forward Deployed Engineer, Platform Governance Product Manager, Product Marketing Manager, Applied AI Engineer, Product Designer, Lead Software Engineer - Growth, Enterprise Account Executive, and Distributed Systems Engineer across San Francisco, Austin, and New York. ([linkedin.com](https://www.linkedin.com/company/arcade-ai/jobs)). Team size observed: 11-50 employees employees.
Source: Arcade.dev LinkedIn jobs page.
https://www.linkedin.com/company/arcade-ai/jobs
-
2026-07-01:
Founding Account Executive joined
(hr)
. Emily Rollo announced she joined Arcade.dev as a Founding Account Executive. ([linkedin.com](https://www.linkedin.com/posts/erollo_im-excited-to-share-that-ive-joined-arcadedev-activity-7467594495174283264-IU6l)). Team size observed: 11-50 employees employees.
Source: Emily Rollo on LinkedIn.
https://www.linkedin.com/posts/erollo_im-excited-to-share-that-ive-joined-arcadedev-activity-7467594495174283264-IU6l
-
2026-06-29:
Cisco Duo step-up launch extended the security ecosystem motion
(gtm)
. Arcade and Cisco Duo were positioned together around content-aware step-up authorization for agent tool calls, showing how Arcade can act at runtime inside enterprise identity and MFA workflows. ([arcade.dev](https://www.arcade.dev/blog/cisco-duo-decides-arcade-acts/))
Source: Cisco Duo Decides, Arcade.dev Acts: Step-Up for Agents.
https://www.arcade.dev/blog/cisco-duo-decides-arcade-acts/
-
2026-06-23:
EMA support for enterprise MCP agents
(product)
. Arcade said it now supports Enterprise-Managed Authorization (EMA), aligning with a newer enterprise MCP authorization pattern and reinforcing its production authorization story. ([arcade.dev](https://www.arcade.dev/blog/arcade-supports-ema/?utm_source=openai))
Source: Arcade.dev blog.
https://www.arcade.dev/blog/arcade-supports-ema/
-
2026-06-23:
EMA support aligned Arcade with identity-provider-managed authorization
(gtm)
. Arcade announced support for Enterprise-Managed Authorization, framing it as validation of its runtime authorization model and inviting early-access interest. ([arcade.dev](https://www.arcade.dev/blog/arcade-supports-ema/))
Source: Arcade.dev Supports EMA for Enterprise MCP Agents.
https://www.arcade.dev/blog/arcade-supports-ema/
-
2026-06-01:
Series A
(funding)
. Amount: 60000000 USD. Investors: SYN Ventures, Morgan Stanley, Wipro
Source: Arcade.dev.
https://www.arcade.dev/blog/arcade-series-a/
Clavitor
Website: clavitor.ai
Clavitor is a credential vault and access-control platform for AI agents that issues scoped credentials, encrypts secrets to the user’s hardware, and logs every access in a tamper-evident audit trail.
Category: Identity, Credentials & Permissions.
Founded: 2026.
Country: United States.
Landscape fit: It fits this category because it gives AI agents a persistent identity, scoped access to external systems, and protected credential handling while emphasizing encryption, auditing, and revocation controls. The product is explicitly built for AI agents and positions itself as a credential vault with per-agent permissions and secure secret injection. Clavitor LLC is listed as a Florida company incorporated on April 1, 2026, which makes 2026 the most supportable founded year from public records. ([clavitor.ai](https://clavitor.ai/en/?utm_source=openai))
Company timeline
-
2026-05-07:
Security-content program expands with DigiCert incident analysis
(gtm)
. Clavitor published another breach-analysis post, this time on DigiCert, continuing a pattern of using high-visibility security incidents to educate prospects and sharpen the company’s threat-model narrative.
Source: Clavitor blog.
https://clavitor.ai/de/blog/digicert-lost-27-code-signing-certificates-to-a-screensaver-file
-
2026-04-27:
Security blog launches breach-driven positioning
(gtm)
. Clavitor published a security-blog post on Vercel’s secrets exposure, using breach analysis to argue for zero-knowledge, hardware-encrypted credential handling and default-safe secret storage for AI-era workflows.
Source: Clavitor blog.
https://clavitor.ai/blog/vercel-stored-your-secrets-in-plaintext-and-called-it-a-feature
Alter
Website: alterauth.com
Alter builds an authorization and access-control layer for AI agents, letting them securely use external APIs with fine-grained permissions, scoped tokens, audit logs, and guardrails.
Category: Identity, Credentials & Permissions.
Funding stage: Pre Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits this category because the product is specifically about persistent agent identity, controlled access to tools and APIs, and secure credential handling for autonomous systems.
Company timeline
-
2026-07-01:
Introduced public usage-based pricing tiers
(gtm)
. Alter published Free, Starter, Growth, and Enterprise plans with API-call limits, overages, audit retention, and support/compliance features.
Source: Alter pricing page.
https://alterauth.com/pricing/
-
2026-07-01:
Rolled out a self-serve early-access product site
(gtm)
. Alter's main site now presents Alter Vault as a productized authorization layer with docs, early access, SDK-driven request flow, Connect/Vault/Policy/Audit surfaces, and 100+ integrations.
Source: Alter website.
https://alterauth.com/
-
2026-04-01:
Published category-setting thesis on agent authorization
(gtm)
. Alter published a long-form post arguing that authorization, not authentication, is the core new problem for AI agents and explaining why traditional machine-to-machine auth breaks down for agentic workflows.
Source: Alter blog.
https://blog.alterauth.com/p/is-authorization-for-ai-agents-actually
-
2025-09-01:
Founder launch of Alter
(hr)
. Co-founders Srikar Dandamuraju and Kevan Dodhia publicly launched Alter, describing the company as an access-control layer for AI agents and asking interested people to reach out.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/kevan-dodhia-760183129_the-last-few-months-have-been-wild-ride-but-activity-7362220279420338179-FsxT
OneCLI
Website: onecli.sh
OneCLI is an open-source credential and policy gateway for AI agents that injects secrets at the network layer, so agents can access external services without ever seeing raw API keys or OAuth tokens.
Category: Identity, Credentials & Permissions.
Funding stage: Seed.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits this category because it gives agents persistent identities, scoped permissions, and secure access to external systems while keeping credentials isolated and revocable.
Company timeline
-
2026-07-01:
Founder-led launch/pivot to OneCLI
(hr)
. Jonathan Haim Fishner publicly described OneCLI as a pivot from prior OSS work and said the team is now building OneCLI full time with Guy Ben-Aharon, backed by Y Combinator. ([linkedin.com](https://www.linkedin.com/company/onecli)). Team size observed: 2-10 employees employees.
Source: LinkedIn / Y Combinator.
https://www.linkedin.com/company/onecli
-
2026-03-30:
Bitwarden integration positioned OneCLI as the runtime enforcement layer
(gtm)
. OneCLI announced Bitwarden's Agent Access SDK integration, pairing approval on the vault side with OneCLI's network-layer credential injection and policy enforcement.
Source: OneCLI blog.
https://onecli.sh/blog/bitwarden-agent-access-sdk-onecli
-
2026-03-30:
Bitwarden integration adds human-approved vault-backed credential flow
(product)
. OneCLI announced an integration with Bitwarden's Agent Access SDK so agents can access vault credentials with human approval while OneCLI handles request-time injection and policy enforcement.
Source: Bitwarden Integrates with OneCLI Agent Vault.
https://onecli.sh/blog/bitwarden-agent-access-sdk-onecli
-
2026-03-25:
NanoClaw adopted OneCLI as its default credential layer
(gtm)
. OneCLI announced that NanoClaw, a widely used open-source agent framework, adopted Agent Vault as its default credential layer.
Source: OneCLI blog.
https://onecli.sh/blog/nanoclaw-adopts-agent-vault-credential-layer
-
2026-03-25:
NanoClaw adopts OneCLI Agent Vault as its default credential layer
(product)
. NanoClaw integrated OneCLI as its default credential layer, using the gateway to proxy requests and inject credentials at the network layer for agent groups.
Source: NanoClaw Now Uses Agent Vault as Its Default Credential Layer.
https://onecli.sh/blog/nanoclaw-adopts-agent-vault-credential-layer
-
2026-03-16:
Rate limiting and approval workflows added to the roadmap
(gtm)
. The company explicitly shifted from pure credential isolation toward policy enforcement, stating that rate limiting and approval workflows were being built for misuse prevention.
Source: OneCLI blog.
https://onecli.sh/blog/credential-vault-ai-agent-security
Agentic Fabriq
Website: agenticfabriq.com
Agentic Fabriq is an identity, authorization, and audit layer for AI agents that gives each agent a verifiable identity, scoped permissions, secure credential handling, and logging across external tools and systems.
Category: Identity, Credentials & Permissions.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits this category because the product is explicitly built to secure AI agents with identities, least-privilege access, credential vaulting, revocation, and audit trails for external systems.
Company timeline
-
2026-07-10:
Formalized production authorization and packaging for adoption
(product)
. Recent materials emphasize agent authorization best practices and the public pricing page now offers Free, Launch, Growth, and Custom tiers with SSO, dedicated realms, custom MCP/API tools, and enterprise controls.
Source: Agentic Fabriq Blog and Pricing.
https://www.agenticfabriq.com/blog/agent-authorization-explained
-
2026-07-03:
Agent operations content program around discovery, visibility, governance, and registry
(gtm)
. In late June and early July 2026, the blog became a coordinated content program about agent discovery, inventory, visibility, registry, governance, lifecycle management, and offboarding, implying a broadened category play toward 'agent operations' rather than just identity. ([agenticfabriq.com](https://www.agenticfabriq.com/blog?utm_source=openai))
Source: Agentic Fabriq blog.
https://www.agenticfabriq.com/blog
-
2026-07-01:
Freemium-to-enterprise pricing and packaging
(gtm)
. The pricing page introduced Free, Launch, Growth, and Custom tiers with usage-based billing, SSO, dedicated realms, custom tools, founder support, dedicated infrastructure, and compliance controls, signaling a more structured sales and packaging motion. ([agenticfabriq.com](https://www.agenticfabriq.com/pricing))
Source: Agentic Fabriq pricing page.
https://www.agenticfabriq.com/pricing
-
2026-06-15:
Expanded the product story to a full governance stack
(product)
. The company began publishing on agent discovery, inventory, registry, visibility, lifecycle management, and offboarding, broadening the product narrative beyond auth into agent operations governance.
Source: Agentic Fabriq Blog.
https://www.agenticfabriq.com/blog/building-an-enterprise-agent-inventory
-
2026-05-11:
Shift from agent building concept to identity infrastructure framing
(gtm)
. In January and May 2026, the messaging moved from a broader no-code/low-code agent service idea to a more explicit identity layer: agent identity as a verifiable principal, with scopes, delegation, audit, and revocation at the center. ([agenticfabriq.com](https://www.agenticfabriq.com/blog/llm-no-code-agents))
Source: Agentic Fabriq blog.
https://www.agenticfabriq.com/blog/identity-for-ai-agents-primer
-
2026-05-11:
Shifted toward identity and permissioning for AI agents
(product)
. The company published a technical primer and related posts explaining agent identity, OAuth/MCP permissioning, and the need for a separate authorization layer for agents.
Source: Agentic Fabriq Blog.
https://www.agenticfabriq.com/blog/identity-for-ai-agents-primer
Payments & Financial Accounts
Infrastructure that allows agents to transact autonomously while remaining within predefined financial controls.
Sponge
Website: paysponge.com
Sponge builds financial infrastructure for AI agents, giving them wallets, bank accounts, cards, and payment rails so they can transact autonomously and businesses can sell directly to agents.
Category: Payments & Financial Accounts.
Funding stage: Seed.
Founded: 2026.
Country: United States.
Employees: 3 employees.
Landscape fit: It fits Payments & Financial Accounts because its core product lets AI agents hold funds, make purchases, and access payment rails within predefined controls rather than just analyze or route payments.
Company timeline
-
2026-07-17:
Gateway positioning shifted Sponge toward merchant API monetization
(gtm)
. Sponge's Gateway pages repositioned the company around turning any API into a paid endpoint, with x402 and MPP support, zero code changes, and public catalog pages for services.
Source: Sponge Gateway.
https://paysponge.com/gateway
-
2026-07-01:
Gateway launched as a paid API monetization layer
(product)
. Sponge Gateway now sits in front of upstream APIs and makes routes payable via x402 or MPP, with OpenAPI import, endpoint pricing, public catalog pages, custom slugs, and MCP control.
Source: Sponge Gateway.
https://paysponge.com/gateway
-
2026-06-01:
Catalog docs introduced a discover-and-pay API layer
(product)
. Sponge Catalog documentation added a separate product for discovering paid APIs and calling them with wallet-based payments through an agent skill or REST API.
Source: Sponge Catalog Docs.
https://catalog.paysponge.com/docs
-
2026-06-01:
Catalog launched as a paid-API discovery layer for agents
(gtm)
. Sponge introduced a catalog where agents can browse pay-per-use APIs, inspect pricing metadata, and make paid calls through x402 fetch.
Source: Sponge Catalog.
https://catalog.paysponge.com/
-
2026-05-01:
Wallet MCP exposed a broader toolset for autonomous agents
(product)
. The wallet MCP surface documented balances, transfers, swaps, paid fetch, x402/MPP fetch, key storage, card storage, approvals, and trading operations for agent frameworks.
Source: Sponge Wallet MCP Docs.
https://docs.paysponge.com/wallet/mcp
-
2026-05-01:
Gateway docs added merchant workflow and pricing concepts
(product)
. Gateway documentation now details service creation, endpoint pricing, upstream auth methods, session flows, and public sharing rules for paid APIs.
Source: Sponge Gateway Docs.
https://docs.paysponge.com/gateway
Locus
Website: paywithlocus.com
Locus builds payment infrastructure for AI agents, giving agents a USDC wallet, spending controls, audit logs, and pay-per-use access to APIs, checkout, and other services.
Category: Payments & Financial Accounts.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Payments & Financial Accounts because its core product lets AI agents hold funds, spend under preset controls, and transact with external services and merchants while remaining governed and auditable.
Company timeline
-
2026-07-23:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/paywithlocus
-
2026-07-01:
Beta repositioned as the main home for build, checkout, and catalog
(product)
. The homepage stated that Build with Locus, Checkout with Locus, and the full API catalog are available exclusively at beta.paywithlocus.com.
Source: Locus website.
https://paywithlocus.com/
-
2026-07-01:
Endpoint catalog expanded to dozens of providers
(product)
. The public endpoints catalog showed a much broader pay-per-use network, listing 44 providers and 33 endpoints also live on MPP.
Source: Locus website.
https://paywithlocus.com/endpoints
-
2026-07-01:
Agent-to-agent payments use case added
(product)
. Locus published an agent-to-agent payments page describing per-agent wallets, budgets, and multi-agent swarm payment flows.
Source: Locus website.
https://paywithlocus.com/use-cases/agent-to-agent-payments
-
2026-07-01:
MCP payments surfaced as a new agent-tool payment layer
(product)
. Locus added MCP payments messaging showing an MCP server that exposes wrapped APIs as tools with built-in payment handling.
Source: Locus website.
https://paywithlocus.com/use-cases/mcp-payments
-
2026-07-01:
Formalized pricing around per-call usage and provider breadth
(gtm)
. The pricing page emphasized per-call billing, one-wallet access, no subscriptions, and a broad endpoint catalog spanning models, search, data, and developer tools. ([paywithlocus.com](https://paywithlocus.com/pricing?utm_source=openai))
Source: Locus Pricing and Endpoints.
https://paywithlocus.com/pricing
Skyfire
Website: skyfire.xyz
Skyfire is an agentic commerce platform that provides identity and payment infrastructure so AI agents can authenticate, make and receive payments, hold balances, and access services autonomously. ([skyfire.xyz](https://skyfire.xyz/about))
Category: Payments & Financial Accounts.
Funding stage: Seed.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: Skyfire provides identity-linked payment infrastructure that lets AI agents pay for digital services and transact autonomously.
Funding rounds
-
Seed extension
announced 2024-10-01
: 1000000 USD
; investors: Coinbase Ventures, a16z Crypto Startup Accelerator
; source: Business Wire
; https://www.businesswire.com/news/home/20241024532897/en/AI-Agents-Race-to-Join-Skyfire-Payments-Network
-
Seed
announced 2024-08-01
: 8500000 USD
; investors: Neuberger Berman, Inception Capital, Arrington Capital, Brevan Howard Digital, Intersection Growth Partners, DRW, RedBeard Ventures, Sfermion, Circle, FBG, Gemini, Crossbeam Venture Partners, EveryRealm, Draper Associates, ARCA, Ripple
; source: TechCrunch
; https://techcrunch.com/2024/08/21/skyfire-lets-ai-agents-spend-your-money/
Company timeline
-
2026-05-01:
KYAPay token model clarified across checkout, subscriptions, and protected access
(product)
. Skyfire’s token documentation now frames KYAPay as a single interoperable credential for account creation, ecommerce checkout, micropayments, subscriptions, and protected website access.
Source: Skyfire Docs.
https://docs.skyfire.xyz/docs/kyapay-tokens
-
2026-05-01:
Jon Fahrner joined as VP of Merchant Development and Services
(hr)
. Skyfire announced that Jon Fahrner joined the team as Vice President of Merchant Development and Services.. Team size observed: 11-50 employees employees.
Source: LinkedIn company page.
https://www.linkedin.com/company/tryskyfire
-
2026-03-01:
Service directory and verified seller trust layer
(product)
. Skyfire introduced an open service directory for agent-discoverable sellers and a verified seller model tied to KYA verification and a dispute-backed service guarantee.
Source: Skyfire Docs.
https://docs.skyfire.xyz/docs/service-discovery
-
2026-02-11:
Agent tool-discovery narrative expands
(gtm)
. Skyfire’s February 2026 technical post on agents dynamically installing tools introduced Skyfire’s directory as part of the flow, suggesting an expanded story around discovery, access, and tool distribution inside agentic commerce. ([skyfire.xyz](https://skyfire.xyz/from-discovery-to-execution-how-ai-agents-can-dynamically-install-tools/?utm_source=openai))
Source: Skyfire blog.
https://skyfire.xyz/from-discovery-to-execution-how-ai-agents-can-dynamically-install-tools/
-
2026-01-11:
Onboarding and KYC education series
(gtm)
. Skyfire published a set of 2026 technical posts focused on identity verification during onboarding, payment automation under surge loads, and payment API resilience, signaling a more explicit education program around operational trust and enterprise readiness. ([skyfire.xyz](https://skyfire.xyz/ai-identity-verification-slows-down-kyc-during-onboarding/?utm_source=openai))
Source: Skyfire blog.
https://skyfire.xyz/how-ai-identity-verification-slows-down-kyc-during-onboarding/
-
2025-11-05:
Entered bot-management and web-security conversations with Akamai
(gtm)
. Skyfire hosted a fireside chat with Akamai at the Agentic Commerce event focused on 403 blocks, fraud, and the agent economy.
Source: Skyfire blog.
https://skyfire.xyz/fireside-chat-mr-403-meets-the-agent-economy/
Natural AI, Inc.
Website: natural.com
Natural is a financial technology platform that provides the infrastructure for AI agents to hold accounts, move money, issue cards, and make payments autonomously within defined controls.
Category: Payments & Financial Accounts.
Funding stage: Series A.
Founded: 2025.
Country: United States.
Employees: 17 people.
Landscape fit: It fits Payments & Financial Accounts because its core product is agentic payments infrastructure: wallets, transfers, cards, credit, and controls that let AI agents transact while staying within financial guardrails.
Funding rounds
-
series A
announced 2026-07-01
: 30000000 USD
; investors: Forerunner, Human Capital, Abstract, Genius Ventures, Torch Capital, Restive, SV Angel, Darragh Buckley, Pablo Palafox, Paul Klein IV, Akshay Kothari, Henri Stern, Max Segall, Pete Kooman, Dylan Babbs, Art Levy, Nichole Wischoff, Jake Paul, Logan Paul
; source: Natural
; https://www.natural.com/blog/natural-series-a
-
seed
announced 2025-10-01
: 9800000 USD
; investors: Abstract, Human Capital, Forerunner, Terrain, Restive, Genius Ventures, Zach Abrams, Immad Akhund, Eric Glyman, Karim Atiyeh, Guillermo Rauch, Itai Damti, Matt Michaelis
; source: Natural
; https://www.natural.com/blog/natural-seed-round
Company timeline
-
2026-07-20:
Series A post confirms team has grown to 17 and hiring is active
(hr)
. Natural’s Series A announcement says the company has a team of 17 and is hiring aggressively across every function. ([natural.com](https://www.natural.com/blog/natural-series-a)). Team size observed: 17 people employees.
Source: Natural Blog.
https://www.natural.com/blog/natural-series-a
-
2026-07-20:
Series A announcement shows broadened platform and platform-sales motion
(gtm)
. Natural announced a $30M Series A and described a wider live product stack including Wallets, Vaults, Pay, Request, Transfer, and Connect, with Cards, Accept, Credit, Charge, Voice, and Billing rolling out or upcoming.
Source: Natural blog.
https://www.natural.com/blog/natural-series-a
-
2026-07-20:
Expanded from core payments into a broader live platform stack
(product)
. Natural's Series A announcement stated that six products were live and the surface now covered Wallets, Vaults, Pay, Request, Transfer, and Connect, with Accept, Cards, Voice, Charge, Billing, Credit, and Direct also exposed as part of the roadmap and product family.
Source: Natural.
https://www.natural.com/blog/natural-series-a
-
2026-07-01:
series A
(funding)
. Amount: 30000000 USD. Investors: Forerunner, Human Capital, Abstract, Genius Ventures, Torch Capital, Restive, SV Angel, Darragh Buckley, Pablo Palafox, Paul Klein IV, Akshay Kothari, Henri Stern, Max Segall, Pete Kooman, Dylan Babbs, Art Levy, Nichole Wischoff, Jake Paul, Logan Paul
Source: Natural.
https://www.natural.com/blog/natural-series-a
-
2026-04-21:
CLI joins as Chief Compliance and Regulatory Officer
(hr)
. Natural announced CLI as Chief Compliance and Regulatory Officer; the post says she was the first regulatory hire at Square, Stripe, and Ramp. ([natural.com](https://www.natural.com/blog/cli-joins-natural?utm_source=openai)). Team size observed: 17 people employees.
Source: Natural Blog.
https://www.natural.com/blog/cli-joins-natural
-
2026-04-21:
Compliance leadership added to support regulated scale
(gtm)
. Natural brought on a Chief Compliance and Regulatory Officer and explicitly framed the hire as building the compliance foundation for agentic payments.
Source: Natural blog.
https://www.natural.com/blog/cli-joins-natural
Allowance
Website: useallowance.com
Allowance is a payment-security layer for AI agents that lets users approve and control agent purchases with one-time virtual cards, spending limits, merchant restrictions, and expiry rules.
Category: Payments & Financial Accounts.
Founded: 2026.
Country: United States.
Employees: 1.
Landscape fit: It fits the Payments & Financial Accounts category because it provides the transaction authorization, card provisioning, and spend-control infrastructure that lets AI agents pay while staying inside predefined financial controls.
Company timeline
-
2026-07-16:
Launched ChatGPT plugin
(gtm)
. Allowance launched on ChatGPT, giving ChatGPT Work and Codex a direct way to request approval and complete purchases without seeing the user’s real card.
Source: Allowance blog.
https://useallowance.com/blog/allowance-plugin-launches-on-chatgpt/
-
2026-07-16:
Allowance plugin launched on ChatGPT
(product)
. Allowance launched as a ChatGPT plugin, extending secure purchase approvals to ChatGPT, ChatGPT Work, and Codex.
Source: Allowance blog.
https://useallowance.com/blog/allowance-plugin-launches-on-chatgpt/
-
2026-07-09:
Expanded distribution through Claude connector
(gtm)
. Allowance became an official connector in Claude, letting users approve purchases on their phone while Claude completes checkout with a scoped one-time card.
Source: Allowance blog.
https://useallowance.com/blog/allowance-official-connector-claude/
-
2026-07-09:
Allowance became an official connector in Claude
(product)
. Allowance added direct support in Claude's Connectors Directory so Claude users can approve purchases and receive scoped one-time cards without MCP setup.
Source: Allowance blog.
https://useallowance.com/blog/allowance-official-connector-claude/
-
2026-05-22:
Public launch on Launch YC
(gtm)
. Allowance launched as a one-time virtual card layer for AI agents, positioning the product around safe checkout with approval controls and built-in spending limits.
Source: Allowance blog / Launch YC.
https://useallowance.com/blog/allowance-virtual-cards-for-ai-agents/
-
2026-05-22:
Public launch of Allowance's virtual-card agent wallet
(product)
. Allowance launched a public product that gives AI agents one-time virtual cards with spending limits, merchant restrictions, approvals, and expiry rules.
Source: Allowance blog.
https://useallowance.com/blog/allowance-virtual-cards-for-ai-agents/
AgentCard
Website: agentcard.sh
AgentCard provides virtual cards and payment controls so AI agents can make online purchases autonomously with predefined spend limits and real-time transaction tracking.
Category: Payments & Financial Accounts.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Payments & Financial Accounts because it gives AI agents a controlled way to hold and spend funds, issue cards, and transact online while staying within preset financial limits.
Company timeline
-
2026-07-20:
Company and personal products consolidated into a broader platform
(gtm)
. Agentcard merged its CLIs, exposed a shell-based API namespace, and introduced cross-app wallet views plus own-card attachment support.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
-
2026-07-15:
Withdrawals and identity-verification sandbox tooling
(product)
. Agentcard added bank and wire withdrawals for organizations and individuals, plus new sandbox tools to test identity verification end to end.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
-
2026-07-13:
Company wallet controls and branded integration flow
(product)
. Agentcard added organization markups on transactions, branded email for the connect flow, and a move away from API keys in the dashboard playground.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
-
2026-07-09:
Organization OAuth2 and MCP-only integration
(product)
. Agentcard added OAuth2 client credentials for organizations, MCP-only integration across cardholders and wallets, phone verification and wallet funding in-conversation, and a Live/Sandbox toggle.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
-
2026-06-18:
Hosted merchant connection and general availability
(product)
. Agentcard added hosted merchant login, order tracking, delivery options, and one free real order for free plans while removing beta and waitlist wording.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
-
2026-06-17:
Conversational buying workflow and Pro plan
(product)
. Agentcard launched a new `buy` tool that runs conversational purchasing end to end and introduced a Pro plan with higher card limits and per-card spend caps.
Source: Agentcard Updates.
https://www.agentcard.sh/updates
AgentCash
Website: agentcash.dev
AgentCash is a CLI, MCP server, and agent wallet that lets AI agents discover and pay for premium APIs and paywalled services using USDC micropayments.
Category: Payments & Financial Accounts.
Funding stage: Bootstrapped.
Founded: 2026.
Country: unknown.
Landscape fit: It fits Payments & Financial Accounts because it provides the financial rail and wallet layer that lets agents make controlled autonomous payments for APIs and services, using USDC and payment-protected endpoints.
Company timeline
-
2026-07-14:
Educational category marketing expanded
(gtm)
. AgentCash published a plain-language agentic-commerce guide to explain stablecoin settlement, x402, MPP, and how the product fits into the stack. ([agentcash.dev](https://agentcash.dev/blog/agentic-commerce-for-dummies?utm_source=openai))
Source: AgentCash blog.
https://agentcash.dev/blog/agentic-commerce-for-dummies
-
2026-07-13:
Provider-side sell-to-agents motion launched
(gtm)
. AgentCash published a merchant playbook and router positioning that tells API providers how to make their services discoverable and payable by agents using x402 and MPP. ([agentcash.dev](https://agentcash.dev/docs/sell-to-agents?utm_source=openai))
Source: AgentCash blog.
https://agentcash.dev/blog/sell-to-agents
-
2026-07-13:
Open-sourced the AgentCash router for merchants
(product)
. AgentCash announced the open-source router as a one-shot way to wrap APIs in x402 and MPP, with discovery docs and wallet-based identity for agents.
Source: AgentCash Blog.
https://agentcash.dev/blog/sell-to-agents
-
2026-07-07:
AgentCash launched as a wallet for AI agents
(product)
. AgentCash introduced a wallet that lets AI agents pay for premium, paywalled APIs and services with one balance and no per-vendor signup.
Source: AgentCash Blog.
https://agentcash.dev/blog/introducing-agent-cash
-
2026-07-01:
Expanded into CLI, MCP, wallet, and discovery workflow
(product)
. Official docs and site messaging added the CLI as the primary interface, MCP server mode, wallet management, API discovery, and paid fetch/discover commands.
Source: AgentCash Docs.
https://agentcash.dev/docs
-
2026-05-01:
MCP and CLI distribution motion hardened
(gtm)
. AgentCash documented CLI and MCP server installation as the primary self-serve ways to onboard agents into paid API access. ([agentcash.dev](https://agentcash.dev/docs/cli/overview?utm_source=openai))
Source: AgentCash docs.
https://agentcash.dev/docs/cli/overview
Roving
Website: roving.money
Roving is a London-based fintech building banking infrastructure for AI agents: it offers agent bank accounts, virtual cards, global payments, spending rules, approvals, and audit trails so agents can transact under human-defined controls. ([roving.money](https://www.roving.money/?utm_source=openai))
Category: Payments & Financial Accounts.
Founded: 2026.
Country: United Kingdom.
Landscape fit: It fits Payments & Financial Accounts because the product is explicitly built to let AI agents hold funds, receive cards, and move money while staying within predefined limits, approvals, permissions, and audit controls. ([roving.money](https://www.roving.money/?utm_source=openai))
AgentWallet
Website: agentwallet.ai
AgentWallet is a financial identity and payments platform for AI agents that provides each agent with a wallet, card, payouts, inbox/phone identity, and policy controls so it can spend and receive money autonomously within approved limits.
Category: Payments & Financial Accounts.
Founded: 2022.
Country: Israel.
Landscape fit: It fits this category because it is explicitly built for agentic commerce: the product lets AI agents hold funds, pay vendors, receive payouts, and transact under predefined caps, approvals, and audit trails.
Company timeline
-
2026-07-01:
Careers page shows 7 open roles
(hr)
. AgentWallet's careers page says the company is hiring across engineering, design, and go-to-market, with 7 open roles listed.
Source: AgentWallet Careers.
https://agentwallet.ai/careers/
-
2026-05-04:
Master technical reference v0.5
(product)
. AgentWallet published a full rewrite of the technical reference covering all shipped subsystems as the public source of truth.
Source: AgentWallet Changelog.
https://agentwallet.ai/changelog/
-
2026-05-01:
Public SEO and route recovery work
(gtm)
. AgentWallet recovered its SEO surface with pre-rendered routes, per-page metadata, and legacy redirects across the marketing site. ([agentwallet.ai](https://agentwallet.ai/changelog?utm_source=openai))
Source: AgentWallet changelog.
https://agentwallet.ai/changelog
-
2026-04-28:
Per-agent x402 wallet auto-topup
(product)
. AgentWallet added automatic top-ups for agent USDC wallets when balances fall below a threshold, running inside the x402 send path.
Source: AgentWallet Changelog.
https://agentwallet.ai/changelog/
-
2026-04-19:
AP2 mandate audit and strict algorithm checks
(product)
. AgentWallet added an AP2 mandate audit table and pinned JWT verification to ES256 to block algorithm confusion attacks.
Source: AgentWallet Changelog.
https://agentwallet.ai/changelog/
-
2026-04-09:
Unified activity feed in the Agent Portal
(product)
. AgentWallet shipped a single portal timeline combining ledger, payouts, card transactions, inbox, outbox, MCP tool calls, and mandate audit events.
Source: AgentWallet Changelog.
https://agentwallet.ai/changelog/
Prava Payments
Website: prava.space
Prava builds payments infrastructure for AI agents, letting AI applications authorize, control, and complete purchases with built-in spend limits, approvals, and card-network-based safeguards.
Category: Payments & Financial Accounts.
Founded: 2025.
Country: United States.
Employees: 2-10 employees (4 visible on LinkedIn).
Landscape fit: It fits the category because it is explicitly a payments and financial-controls layer for AI agents, enabling autonomous transactions under user-defined limits and approvals.
Company timeline
-
2026-07-01:
Small team visible on LinkedIn
(hr)
. Prava Payments' LinkedIn company page currently shows a 2-10 employee range and four visible employees.. Team size observed: 2-10 employees (4 visible on LinkedIn) employees.
Source: Prava Payments LinkedIn.
https://www.linkedin.com/company/pravapayments
-
2026-03-11:
Shifted to multi-rail payment routing
(gtm)
. Prava argued that cards, stablecoins, and ACH each win different transaction types and said its system routes to the optimal rail based on the transaction.
Source: Prava blog.
https://www.prava.space/blog/why-the-cards-vs.-crypto-debate-for-ai-agents-is-a-false-binary
-
2026-02-25:
Announced Visa Intelligent Commerce integration
(gtm)
. Prava announced an integration with Visa Intelligent Commerce to enable secure, card-based agentic payments across AI shopping, travel, and personal-assistant platforms in the US.
Source: Prava blog.
https://www.prava.space/blog/prava-integrates-visa-intelligent-commerce-to-enable-secure-card-based-agentic-payments-across-ai-applications
-
2026-02-25:
Visa Intelligent Commerce integration launched
(product)
. Prava announced an integration with Visa Intelligent Commerce to enable secure, card-based agentic payments for AI shopping, travel, and personal assistant applications in the United States.
Source: Prava blog.
https://www.prava.space/blog/prava-integrates-visa-intelligent-commerce-to-enable-secure-card-based-agentic-payments-across-ai-applications
-
2026-01-15:
Added protocol-map thought leadership and rail-agnostic framing
(gtm)
. Prava published a protocol landscape article that positioned the company as building a processor-agnostic, protocol-agnostic, rail-agnostic stack for agentic commerce.
Source: Prava blog.
https://www.prava.space/blog/the-agentic-payments-map-every-protocol-standard-and-player-you-need-to-know
-
2025-12-30:
Expanded use-case narrative into travel
(gtm)
. Prava published a travel-focused piece arguing that multi-leg booking and multi-merchant orchestration are a key agentic-payments use case.
Source: Prava blog.
https://www.prava.space/blog/ai-travel-agents-are-coming-back.-this-time-they-ll-actually-book-the-trip.
Orthogonal
Website: orthogonal.com
Orthogonal is a unified API and payments platform that lets AI agents discover, access, authenticate with, and pay for third-party APIs and services in one workflow. ([orthogonal.com](https://www.orthogonal.com/terms?utm_source=openai))
Category: Payments & Financial Accounts.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Payments & Financial Accounts because its core product is agentic payment infrastructure: agents can discover services, use managed credentials, and pay per call within predefined controls and auditable billing. ([orthogonal.com](https://www.orthogonal.com/about/?utm_source=openai))
Funding rounds
-
Seed
announced 2026-06-01
: 4300000 USD
; investors: Pantera Capital, Y Combinator, Pioneer Fund, Decasonic, Blast, Outbound, Surreal Ventures, Rice Capital, Batch Ventures
; source: Orthogonal
; https://www.orthogonal.com/blog/orthogonal-raises-4-3-million-seed-round
Company timeline
-
2026-07-16:
Continued catalog expansion into contact and company data
(gtm)
. Orthogonal added RocketReach as a partner spotlight, giving agents access to professional profiles, companies, and contact enrichment through one integration.
Source: Orthogonal Blog.
https://www.orthogonal.com/blog
-
2026-07-01:
Hiring first Founding Engineer announced on LinkedIn
(hr)
. Orthogonal’s LinkedIn company page posts that the company is hiring its first Founding Engineer and links applicants to the careers page.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/orthogonalsh
-
2026-07-01:
General application careers page active
(hr)
. Orthogonal maintains a live general application page inviting candidates to apply even when they do not match an open role.. Team size observed: 2-10 employees employees.
Source: Orthogonal.
https://www.orthogonal.com/careers/general
-
2026-07-01:
Opened Founding Engineer role
(hr)
. Orthogonal posted a Founding Engineer job for San Francisco Bay Area, describing it as one of the company’s first hires and reporting to the founders directly.. Team size observed: 2-10 employees employees.
Source: Orthogonal.
https://www.orthogonal.com/careers/founding-engineer
-
2026-07-01:
Founder-led company site and team page published
(hr)
. Orthogonal’s about page publicly identifies co-founders Christian Pickett and Bera Sogut and describes the company as founded by a payments/API/infrastructure team.. Team size observed: 2-10 employees employees.
Source: Orthogonal.
https://www.orthogonal.com/about/
-
2026-06-25:
Raised seed funding with evidence of production usage
(gtm)
. Orthogonal announced a $4.3 million seed round and said it was already live in production with GTM teams and AI-native startups, hosting more than 35 APIs.
Source: Orthogonal Blog.
https://www.orthogonal.com/blog/orthogonal-raises-4-3-million-seed-round
Kite
Website: gokite.ai
Kite is building payment infrastructure for the agent-native web, giving autonomous AI agents verifiable identity, programmable governance, and native stablecoin settlement.
Category: Payments & Financial Accounts.
Funding stage: Series A.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: Kite gives AI agents identity-linked payment infrastructure for programmable transactions and settlement.
Funding rounds
-
Series A
announced 2025-09-01
: 18000000 USD
; investors: PayPal Ventures, General Catalyst, 8VC, Samsung Next, Alumni Ventures, SBI US Gateway Fund, Vertex Ventures, Avalanche Foundation, GSR, Animoca Brands, Hashed, HashKey Capital
; source: Fortune
; https://fortune.com/crypto/2025/09/02/kite-ai-blockchain-paypal-general-catalyst-18-million/
-
Seed
announced 2025-02-01
: 15000000 USD
; investors: General Catalyst, Hashed, HashKey Capital
; source: CoinCarp
; https://www.coincarp.com/fundraising/kite-ai-seed/
Company timeline
-
2026-05-01:
Mainnet and live product GTM
(gtm)
. Kite’s homepage now says mainnet is live and directs users to try Kite Agent Passport and build on Kite Chain.
Source: Kite - The Payments Layer for the Agent Economy.
https://gokite.ai/
-
2026-05-01:
Hiring push across core functions
(hr)
. Kite posted that it is hiring in its San Francisco office across engineering, AI research, protocol, product, design, and growth, and directed candidates to open roles.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/gokiteai
-
2026-04-01:
Open product and infrastructure roles in San Francisco
(hr)
. Kite posted hiring for Head of Product, Lead Backend Engineer, Product Designer, and Blockchain Infra Engineer, all framed as San Francisco Bay Area in-person roles. Together, these postings signal a concentrated hiring cluster to build out product, backend, and design around the core platform. ([linkedin.com](https://www.linkedin.com/jobs/view/head-of-product-at-kite-4326366938?utm_source=openai))
Source: LinkedIn jobs.
https://www.linkedin.com/jobs/view/head-of-product-at-kite-4326366938
-
2026-04-01:
Hackathon pipeline and community recruiting expanded
(hr)
. Kite launched and promoted the Kite AI Global Hackathon 2026 with partners including Google Cloud, Coinbase Ventures, Avalanche, and UC Berkeley groups. This looks like deliberate talent-brand and pipeline-building aimed at builders, students, and ecosystem contributors. ([linkedin.com](https://www.linkedin.com/posts/gokiteai_thrilled-to-announce-our-first-strategic-activity-7442989822824095744--qeG?utm_source=openai))
Source: Kite LinkedIn.
https://www.linkedin.com/company/gokiteai
-
2026-02-01:
Founder-led talent branding around the team
(hr)
. Kite’s founders publicly discussed how Kite is built, including Scott Shi describing the company as a response to infrastructure gaps for agents and emphasizing lessons from prior teams. This is a clear founder-led narrative used to attract builders and explain the mission to candidates. ([linkedin.com](https://www.linkedin.com/posts/scottrshi_when-sam-altman-said-in-ai-startup-school-activity-7368829044965994498-bKtr?utm_source=openai))
Source: Kite LinkedIn.
https://www.linkedin.com/posts/scottrshi_when-sam-altman-said-in-ai-startup-school-activity-7368829044965994498-bKtr
-
2025-11-21:
OKX Wallet community partnership
(gtm)
. OKX Wallet announced Kite as a community partner to help users access the Kite ecosystem and AI-powered payment experiences.
Source: Kite AI Partners with OKX Wallet to Accelerate AI-Powered Payments.
https://medium.com/%40okxweb3/kite-ai-partners-with-okx-wallet-to-accelerate-ai-powered-payments-acbb4ae71897
Catena
Website: catenalabs.com
Catena is building a governance and banking platform for AI agents, combining programmable financial infrastructure, policy controls, identity, and audit trails so agents can transact safely.
Category: Payments & Financial Accounts.
Funding stage: Series A.
Founded: 2022.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits this category because it provides the financial rails, controls, and account infrastructure needed for AI agents to hold money and make payments within predefined limits.
Funding rounds
-
Seed
announced 2025-05-01
: 18000000 USD
; investors: a16z crypto, Breyer Capital, Circle Ventures, Coinbase Ventures, CoinFund, Pillar VC, Stanford Engineering VF
; source: Catena Labs
; https://catenalabs.com/blog/announcing-catena-labs
Company timeline
-
2026-02-06:
Released a real-money agent treasury demo
(gtm)
. Catena published a demo showing how agents need guardrails before handling real money and framed the work as a response to wider awareness of agent risk.
Source: Catena Labs.
https://catenalabs.com/blog/what-agents-need-before-they-handle-real-money
-
2026-02-06:
Showed real-money agent treasury and payments demo
(product)
. Catena published a raw demo of agent treasury management and payments with identity and policy guardrails, including agent hiring, approvals, rejections, and stablecoin-based funds handling.
Source: Catena Labs blog.
https://catenalabs.com/blog/what-agents-need-before-they-handle-real-money
-
2025-12-18:
Began using its own agent banking system on live corporate funds
(gtm)
. Catena said its AI agent banking system started managing a portion of its own corporate funds, presenting the company as Client Zero.
Source: Catena Labs.
https://catenalabs.com/blog/being-client-zero
-
2025-12-18:
Started using its AI agent banking system on internal corporate funds
(product)
. Catena said its AI agent banking system began managing a portion of its own corporate funds with live agents and policy boundaries.
Source: Catena Labs blog.
https://catenalabs.com/blog/being-client-zero
-
2025-10-28:
Partnered with Circle to support Arc
(gtm)
. Catena announced a partnership with Circle to build support for agentic commerce and finance on Arc, Circle’s payments and stablecoin-focused Layer-1 blockchain.
Source: Catena Labs.
https://catenalabs.com/blog/catena-partners-with-circle-to-support-arc-built-for-agentic-finance-and-commerce
-
2025-08-13:
Clarified the agentic finance stack and bank layer focus
(gtm)
. Catena published a detailed framework for the agentic finance stack, emphasizing that it is building at the banking and financial-services layer for AI agents.
Source: Catena Labs.
https://catenalabs.com/blog/the-a-commerce-stack-building-the-financial-capabilities-for-ai-agents
Nevermined
Website: nevermined.ai
Nevermined builds payments and financial infrastructure for AI agents, enabling agent-to-agent commerce, real-time metering, and autonomous pay/get-paid workflows across cards, fiat, and crypto.
Category: Payments & Financial Accounts.
Funding stage: Seed.
Founded: 2022.
Country: Switzerland.
Employees: 2-10 employees.
Landscape fit: Nevermined gives AI agents payment and settlement infrastructure for autonomous commerce and pay-per-use services.
Funding rounds
-
Seed
announced 2025-01-01
: 4000000 USD
; investors: Generative Ventures, Polymorphic Capital, NEAR, Halo Capital, Factor Capital, Lyrik Ventures, Arca
; source: GlobeNewswire
; https://www.globenewswire.com/news-release/2025/01/09/3007256/0/en/Nevermined-Raises-4M-to-Build-the-PayPal-for-AI-Commerce.html
-
Extended Seed
announced 2025-01-01
: $4M USD
; investors: Generative Ventures, Polymorphic Capital, NEAR, Halo Capital, Factor Capital, Lyrik Ventures, Arca
; source: GlobeNewswire
; https://www.globenewswire.com/news-release/2025/01/09/3007256/0/en/Nevermined-Raises-4M-to-Build-the-PayPal-for-AI-Commerce.html/
-
Seed
announced 2022-07-01
: 3000000 EUR
; investors: Signature Ventures, Polymorphic Capital
; source: Business Wire
; https://www.businesswire.com/news/home/20220712005503/en/Nevermined-Raises-%E2%82%AC3-Million-Seed-Funding-to-Make-Web3-Development-Easier-for-Creators-and-Developers
Company timeline
-
2026-05-06:
Exa partnership showcases autonomous agent spending in a real workflow
(gtm)
. Nevermined highlighted Exa as a partner example for AI agents making autonomous payments, reinforcing the merchant/consumer-side use case with a recognizable AI search company.
Source: Nevermined blog.
https://nevermined.ai/blog/exa-nevermined-autonomous-ai-agent-payments
-
2026-05-06:
Exa enabled autonomous agent payments through Nevermined
(product)
. Nevermined announced that Exa now accepts autonomous payments from AI agents using delegated credit cards via its x402 card-delegation flow, with instant API key issuance after purchase.
Source: Nevermined blog.
https://nevermined.ai/blog/exa-nevermined-autonomous-ai-agent-payments
-
2026-05-01:
Developer Advocate role added for builder-facing growth
(hr)
. Nevermined is hiring a Developer Advocate for AI + Payments in San Francisco to write guides, host workshops, and ship sample apps for agent monetization.
Source: Nevermined About Us.
https://nevermined.ai/about-us
-
2026-05-01:
Careers page shows active hiring for Developer Advocate - AI + Payments
(hr)
. Nevermined’s About/Team page includes a live hiring section and a listed opening for Developer Advocate - AI + Payments in San Francisco, indicating ongoing hiring for developer-facing support.. Team size observed: 2-10 employees employees.
Source: Nevermined About Us.
https://nevermined.ai/about-us/
-
2026-05-01:
Compact core team is visible on the company site
(hr)
. Nevermined’s about page shows a small founding/core group with one business development hire and a lead engineer, indicating an early-stage, founder-led organization.
Source: Nevermined About Us.
https://nevermined.ai/about-us
-
2026-04-08:
Visa Intelligent Commerce integration opens merchant and publisher enrollment
(gtm)
. Nevermined announced integration with Visa Intelligent Commerce, x402, and VGS to let AI agents buy digital goods and services while merchants accept payments through existing infrastructure.
Source: Nevermined blog.
https://nevermined.ai/blog/nevermined-unlocks-autonomous-agent-card-payments-with-x402-opening-a-new-market-for-publishers-digital-merchants
Rye
Website: rye.com
Rye provides a Universal Checkout API and related commerce infrastructure that lets software, apps, and AI agents discover products, retrieve pricing/shipping/tax, and complete purchases programmatically across merchants.
Category: Payments & Financial Accounts.
Funding stage: Seed.
Founded: 2021.
Country: United States.
Employees: 11-50 employees.
Landscape fit: Rye lets AI agents retrieve checkout details and complete purchases programmatically across online merchants.
Funding rounds
-
Seed
announced 2022-10-01
: 14000000 USD
; investors: a16z crypto, Goat Capital, Solana Ventures, L Catterton, Electric Ant, Electric Feel Ventures, Andre Iguodala, JaVale McGee, James Beshara
; source: Forbes
; https://www.forbes.com/sites/alexkonrad/2022/10/11/a16z-crypto-web3-commerce-rye-justin-kan/
Company timeline
-
2026-05-28:
Current employee-count observation
(hr)
. LinkedIn shows Rye as a 11-50 employee company and also lists 105 employees on the company page.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://at.linkedin.com/company/rye
-
2026-05-12:
Merchant UCP scoping guide positions Rye as the faster integration path
(gtm)
. Rye published a merchant guide comparing native UCP implementation with integrating via Rye, framing Rye as an easier path to agentic-commerce coverage for merchants and platforms. ([rye.com](https://rye.com/blog/how-to-implement-ucp?utm_source=openai))
Source: Rye Blog.
https://rye.com/blog/how-to-implement-ucp
-
2026-04-29:
Added x402 payment support through AgentCash
(gtm)
. Rye announced native x402 support via AgentCash, enabling stablecoin-style payments for physical commerce on top of Universal Checkout.
Source: Rye blog.
https://rye.com/blog/agentcash-x402-agentic-commerce
-
2026-04-29:
x402 physical-commerce support added
(product)
. Rye added native x402 support through AgentCash so agents can pay for physical goods with USDC and complete checkout over HTTP.
Source: Rye blog.
https://rye.com/blog/agentcash-x402-agentic-commerce
-
2026-04-29:
AgentCash x402 support for physical commerce
(product)
. Rye announced that its Universal Checkout API can accept x402 payments via AgentCash, letting agents discover a product, sign a USDC payment, and complete purchases over HTTP.
Source: Rye blog.
https://rye.com/blog/agentcash-x402-agentic-commerce
-
2026-04-21:
Integrated Skyfire identity into checkout stack
(gtm)
. Rye announced a partnership to fold Skyfire's KYA protocol into Universal Checkout as the identity layer for agent traffic.
Source: Rye blog.
https://rye.com/blog/skyfire-agentic-commerce-stack
Communication Channels
Services that give agents their own communication endpoints so they can interact directly with people and other software.
Inkbox
Website: inkbox.ai
Inkbox builds an identity and communication layer for AI agents, giving each agent a persistent inbox, phone number, internet address, and shared context across channels like email, calls, SMS, and iMessage.
Category: Communication Channels.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Communication Channels because it gives agents their own durable endpoints for interacting directly with people and systems across email, phone, iMessage, and web traffic.
Company timeline
-
2026-06-01:
Founders announce Inkbox YC S26 build and move to San Francisco
(hr)
. Ray Liao said he was moving to SF with Alex Wilcox and Dima Vremenko while announcing Inkbox is joining Y Combinator S26. ([linkedin.com](https://www.linkedin.com/posts/ruizhi-liao_every-time-i-visit-a-new-place-i-like-to-activity-7470156972340903937-6mxI?utm_source=openai)). Team size observed: 2-10 employees employees.
Source: LinkedIn posts.
https://www.linkedin.com/posts/ruizhi-liao_every-time-i-visit-a-new-place-i-like-to-activity-7470156972340903937-6mxI
-
2026-04-04:
Added agent self-signup to reduce onboarding friction
(gtm)
. Inkbox launched agent self-signup so an AI agent can create its own identity, mailbox, tunnel, and API key before a human account exists.
Source: Inkbox blog.
https://inkbox.ai/blog
-
2026-04-04:
Agents can self-register for Inkbox
(product)
. Inkbox added agent self-signup, changing auth so an agent can start working before a human account exists. ([inkbox.ai](https://inkbox.ai/blog?utm_source=openai))
Source: Inkbox blog.
https://inkbox.ai/blog
-
2026-01-06:
Open-sourced telephony stack to build infrastructure credibility
(gtm)
. Inkbox publicly framed itself as an open-source, HIPAA-eligible Twilio alternative with production-grade telephony for AI voice agents.
Source: Inkbox blog.
https://inkbox.ai/blog/open-telephony-stack
-
2026-01-06:
Open-source telephony stack positioned as a Twilio alternative
(product)
. Inkbox described itself as an open-source, HIPAA-eligible Twilio alternative built on Asterisk and AWS Chime for production telephony. ([inkbox.ai](https://inkbox.ai/blog?utm_source=openai))
Source: Inkbox blog.
https://inkbox.ai/blog
AgentPost
Website: agentpost.email
AgentPost is an email infrastructure platform for AI agents that lets software create, receive, send, and monitor mailboxes via MCP and API tools.
Category: Communication Channels.
Founded: 2026.
Country: Unknown.
Landscape fit: It fits Communication Channels because it gives agents their own direct email endpoint so they can interact with people and software via inbound/outbound messages.
Company timeline
-
2026-07-01:
Repositioned into a narrower API/MCP email product with waitlist and docs
(gtm)
. The current site focuses on an agent-specific email API that can create, receive, send, and destroy mailboxes via five MCP tool calls, with a waitlist, documentation, and simple tiered pricing for builders and production teams.
Source: agentpost.email.
https://agentpost.email/
-
2026-07-01:
Public launch of MCP-native email infrastructure and pricing tiers
(product)
. The website presents AgentPost as email infrastructure for AI agents with MCP tools to create mailboxes, list messages, read email, send email, and delete mailboxes, alongside a free-to-business pricing ladder.
Source: AgentPost homepage.
https://agentpost.email/
AgentPhone
Website: agentphone.ai
AgentPhone is an API platform that gives AI agents their own phone numbers so they can send SMS/iMessage, receive calls, and hold voice conversations with people and other systems.
Category: Communication Channels.
Funding stage: Pre Seed.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits the Communication Channels category because it gives agents dedicated communication endpoints—phone numbers plus SMS and voice calling—so they can interact directly with humans and software over real-world channels.
Company timeline
-
2026-07-01:
Developer tooling expanded with MCP server and assistant integrations
(product)
. AgentPhone added an MCP server plus integration quickstarts for tools like Claude Code and Google ADK, letting AI assistants buy numbers, send texts, and place calls through natural language.
Source: AgentPhone MCP and Integrations Guides.
https://docs.agentphone.ai/mcp
-
2026-07-01:
Voice calling broadened to hosted AI and streaming web calls
(product)
. AgentPhone documented inbound and outbound calls with real-time transcripts, SSE transcript streaming, and a hosted voice mode that lets the platform run calls end-to-end without a customer webhook.
Source: AgentPhone Calls and Agents Guides.
https://docs.agentphone.ai/documentation/guides/calls
-
2026-07-01:
Unified messaging added SMS, MMS, and iMessage
(product)
. AgentPhone’s messaging API expanded to support SMS, MMS, and iMessage through a single endpoint with automatic channel selection, threading, reactions, effects, and group chats.
Source: AgentPhone Messages Guide.
https://docs.agentphone.ai/documentation/guides/messages
-
2026-07-01:
Hiring first founding engineer
(hr)
. Manav Modi posted that AgentPhone is hiring its first founding engineer and described the role as employee #1 working directly with the founders.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/manav2modi_were-hiring-our-first-founding-engineer-activity-7480758084487917568-_S-G
-
2026-07-01:
Pay-as-you-go billing and usage limits introduced
(product)
. AgentPhone documented usage-based pricing with free starter credit, per-number and per-message fees, call-recording add-ons, denoising surcharges, and starter plan limits.
Source: AgentPhone Usage & Billing.
https://docs.agentphone.ai/usage
-
2026-07-01:
Core agent phone-number API and webhooks documented
(product)
. AgentPhone formalized its core product around creating agents, provisioning SMS- and voice-enabled numbers, attaching numbers to agents, and receiving `agent.message` and `agent.call_ended` webhooks.
Source: AgentPhone Welcome.
https://docs.agentphone.ai/welcome
Primitive
Website: primitive.dev
Primitive builds email and communication infrastructure for AI agents, letting agents send, receive, and manage messages through real email addresses and APIs.
Category: Communication Channels.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Communication Channels because its core product gives agents their own interoperable communication endpoint—email—so they can directly message people and other software without human mailbox access or custom protocols.
Company timeline
-
2026-06-25:
Agent registry and consent standard introduced
(gtm)
. Primitive introduced a public agent registry plus an `_agents` DNS record standard so domains can declare which agents are allowed to contact them. ([primitive.dev](https://www.primitive.dev/blog/launch-week-04?utm_source=openai))
Source: primitive.dev blog.
https://www.primitive.dev/blog/launch-week-04
-
2026-06-25:
Agent registry and agent authorization standard ship
(product)
. Primitive introduced a public registry for agent discovery and proposed the `_agents` DNS record to control which agents may contact a domain.
Source: Primitive blog.
https://www.primitive.dev/blog/launch-week-04
-
2026-06-24:
ChatGPT app launch with inboxed agent workflows
(gtm)
. Primitive launched a ChatGPT app that gives a ChatGPT conversation a real inbox for sending, searching, reading, and replying to email inline. ([primitive.dev](https://www.primitive.dev/blog/launch-week-03?utm_source=openai))
Source: primitive.dev blog.
https://www.primitive.dev/blog/launch-week-03
-
2026-06-24:
ChatGPT app release adds inbox-based agent collaboration
(product)
. Primitive shipped a ChatGPT app experience so a ChatGPT conversation can send, search, read, and reply to email through a real Primitive inbox.
Source: Primitive blog.
https://www.primitive.dev/blog/launch-week-03
-
2026-06-21:
Positioning shifts to agent collaboration over email
(gtm)
. Primitive’s blog reframed the company around agents communicating through ordinary email, arguing that email is the universal layer agents should use instead of closed chat or mailbox workflows. ([primitive.dev](https://www.primitive.dev/blog/it-was-always-email?utm_source=openai))
Source: primitive.dev blog.
https://www.primitive.dev/blog/it-was-always-email
-
2026-06-21:
Public positioning shifts to agent-to-agent email collaboration
(product)
. Primitive published a blog post framing email as the missing collaboration layer for autonomous agents and emphasizing easier cross-agent communication.
Source: Primitive blog.
https://www.primitive.dev/blog/it-was-always-email
AgentMail
Website: agentmail.to
AgentMail builds an API-first email inbox and communication layer designed specifically for AI agents, letting agents send, receive, thread, and manage email like first-class internet citizens.
Category: Communication Channels.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits the Communication Channels category because its core product gives AI agents their own programmable email endpoint for direct two-way interaction with people and software.
Funding rounds
-
seed
announced 2026-03-01
: 6000000 USD
; investors: General Catalyst, Y Combinator, Phosphor Capital, Paul Graham, Taro Fukuyama, Dharmesh Shah, Paul Copplestone, Karim Atiyeh
; source: AgentMail
; https://www.agentmail.to/blog/agentmail-seed-launch
Company timeline
-
2026-07-13:
Native Vercel Marketplace integration
(product)
. AgentMail launched as a native Vercel Marketplace integration so Vercel projects can provision inboxes from the dashboard.
Source: AgentMail Blog.
https://www.agentmail.to/blog/agentmail-on-vercel-marketplace
-
2026-07-13:
Vercel Marketplace integration
(gtm)
. AgentMail launched a native integration on the Vercel Marketplace so deployed projects can provision agent inboxes directly from the Vercel dashboard.
Source: AgentMail blog.
https://www.agentmail.to/blog/agentmail-on-vercel-marketplace
-
2026-07-01:
Current engineering and GTM hiring cluster
(hr)
. AgentMail's careers page currently lists open Engineering roles, including Senior Infrastructure Engineer and Founding Engineer, and a founder post says the company is hiring a GTM Engineer.. Team size observed: 2-10 employees employees.
Source: AgentMail careers page and LinkedIn.
https://www.agentmail.to/careers
-
2026-06-25:
Scoped webhooks and reply/forward drafts
(product)
. AgentMail added pod- and inbox-scoped webhook management plus reply, reply-all, and forward drafts created directly from a message.
Source: AgentMail Changelog.
https://docs.agentmail.to/changelog/2026/6/25
-
2026-06-13:
Official LangChain integration and coding-agent setup guides
(gtm)
. AgentMail published guides for LangChain and coding agents, showing it is packaging the product for popular agent frameworks and coding workflows.
Source: AgentMail blog.
https://www.agentmail.to/blog/give-your-langchain-agent-a-real-inbox
-
2026-06-12:
Usage endpoints split usage from events
(product)
. AgentMail added cumulative usage endpoints for organization, pod, and inbox scopes and moved event counts to dedicated metrics paths.
Source: AgentMail Changelog.
https://docs.agentmail.to/changelog/2026/6/12
Compute & Interaction Environments
Cloud environments that give agents a persistent digital workspace from which they can operate software as a human user would.
Lightpanda
Website: lightpanda.io
Lightpanda builds an open-source headless browser from scratch for machines, with built-in automation and AI-agent workflows, emphasizing speed, low memory use, and browser-programmatic control rather than human-facing rendering. ([lightpanda.io](https://lightpanda.io/))
Category: Compute & Interaction Environments.
Funding stage: Pre Seed.
Founded: 2023.
Country: France.
Employees: 2-10 employees.
Landscape fit: It fits Compute & Interaction Environments because it provides the persistent browser workspace and control layer that AI agents can use to interact with websites programmatically, rather than a consumer browser or general SaaS app. ([lightpanda.io](https://lightpanda.io/))
Funding rounds
-
pre-seed
announced 2025-06-01
: 4500000 USD
; investors: ISAI, Kima Ventures, Factorial Capital, Prototype Capital
; source: Lightpanda
; https://lightpanda.io/blog/posts/lightpanda-raises-preseed
Company timeline
-
2026-07-01:
Careers page shows two senior engineering openings
(hr)
. Lightpanda's careers page currently lists Senior Software Engineer roles for System Programming (Zig/C++) and Web API Implementation (Javascript/Typescript).. Team size observed: 2-10 employees employees.
Source: Lightpanda Careers.
https://lightpanda.io/jobs
-
2026-06-26:
Added agent-browser integration
(product)
. Lightpanda published support for Vercel’s agent-browser CLI, allowing it to run with Lightpanda as the engine through CDP.
Source: Lightpanda blog.
https://lightpanda.io/blog/posts/using-lightpanda-with-agent-browser
-
2026-06-26:
Extends distribution through agent-browser integration
(gtm)
. Lightpanda publishes a guide for using agent-browser with Lightpanda, positioning itself as a backend for AI-agent browser stacks and showing how the engine fits existing workflows.
Source: Lightpanda blog.
https://lightpanda.io/blog/posts/using-lightpanda-with-agent-browser
-
2026-06-17:
Launched Lightpanda Agent and PandaScript
(product)
. Lightpanda introduced a built-in browser agent that turns natural-language tasks into reproducible PandaScript scripts.
Source: Lightpanda blog.
https://lightpanda.io/blog/posts/introducing-lightpanda-agent-and-pandascript
-
2026-06-17:
Launches built-in agent and PandaScript
(gtm)
. Lightpanda introduces a native browser agent and a reproducible scripting layer, framing the product as a browser that can both execute and generate automation workflows.
Source: Lightpanda blog.
https://lightpanda.io/blog/posts/introducing-lightpanda-agent-and-pandascript
-
2026-05-08:
Became an officially supported browser backend in Hermes Agent
(product)
. Lightpanda was added as a browser backend for Nous Research’s Hermes Agent with one-line enablement and Chrome fallback.
Source: Lightpanda blog.
https://lightpanda.io/blog/posts/lightpanda-is-a-browser-backend-hermes-agent
Scrapybara
Website: scrapybara.com
Scrapybara provides virtual desktops and remote compute environments for AI agents to use browsers, code execution, filesystems, and authenticated sessions like a human operator would.
Category: Compute & Interaction Environments.
Funding stage: Seed.
Founded: 2024.
Country: United States.
Employees: 1-10 employees.
Landscape fit: It fits Compute & Interaction Environments because its core product is a persistent cloud desktop that agents can operate interactively, with browser and OS access, session persistence, and orchestration for computer-use workflows.
Company timeline
-
2025-03-11:
OpenAI CUA starter kit and ecosystem integration
(gtm)
. Scrapybara launched an OpenAI CUA starter kit and positioned itself as a production host for OpenAI's computer-use agent, with direct support through its playground, Act SDK, and the Responses API.
Source: Scrapybara Blog.
https://scrapybara.com/blog/cua
-
2025-03-11:
OpenAI CUA starter kit and production integration
(product)
. Scrapybara launched an OpenAI CUA starter kit and documented direct integration paths through its Playground, Act SDK, and OpenAI Responses API.
Source: Scrapybara Blog.
https://scrapybara.com/blog/cua
-
2025-03-07:
Infra and stability messaging for production use
(gtm)
. Scrapybara followed with an infra update post focused on speed, scale, and stability, reinforcing the product as production infrastructure rather than a demo-only tool.
Source: Scrapybara Blog.
https://scrapybara.com/blog
-
2025-03-07:
Infrastructure updates for speed, scale, and stability
(product)
. Scrapybara announced major infrastructure improvements aimed at supporting larger and more performance-intensive computer-use deployments.
Source: Scrapybara Blog.
https://scrapybara.com/blog/speed-scale-stability
-
2025-03-06:
Category positioning around computer-use infrastructure
(gtm)
. Scrapybara published a research post framing its work around a new standard specification for computer-use agent actions, signaling an attempt to shape the category narrative rather than only ship tooling.
Source: Scrapybara Blog.
https://scrapybara.com/blog
-
2025-03-06:
Unified computer-use action space
(product)
. Scrapybara introduced a unified action specification on its computer endpoint so the Act SDK can translate model outputs into a consistent set of mouse, keyboard, and utility actions.
Source: Scrapybara Blog.
https://scrapybara.com/blog/unified-action-space
machine0
Website: machine0.io
machine0 provides persistent cloud VMs that agents can manage from the CLI or MCP, with preinstalled tools, dedicated CPU/RAM, static IPs, and per-minute billing.
Category: Compute & Interaction Environments.
Founded: 2026.
Country: France.
Employees: 2-10 employees.
Landscape fit: It fits this category because it gives agents a durable digital workspace with full VM access, persistence, and programmatic control, which is exactly the kind of compute environment agents need to operate software like a human user would. ([machine0.io](https://machine0.io/))
Company timeline
-
2026-07-21:
Profile variables replaced env sets
(product)
. Machine0 removed the old env-set workflow and moved variable management into profile variables.
Source: machine0 changelog.
https://machine0.io/changelog
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/machine0
-
2026-06-30:
Environment variable sets introduced
(product)
. Machine0 added named env sets that could be attached to new VMs with `machine0 new --env <name>`.
Source: machine0 changelog.
https://machine0.io/changelog
-
2026-05-10:
Redemption-code onboarding launched
(product)
. Machine0 added promo-link onboarding with free credits during sign-up, plus a no-login path for checking sizes and pricing.
Source: machine0 changelog.
https://machine0.io/changelog
-
2026-04-28:
Launch of larger premium CPU tiers
(product)
. Machine0 added 4xl, 5xl, and 6xl premium Intel sizes for much larger dedicated CPU and RAM allocations.
Source: machine0 changelog.
https://machine0.io/changelog
-
2026-04-08:
Core launch bundle expanded with billing and file sync
(product)
. Machine0 added a monthly billing command and `sync push`/`sync pull` file transfer commands with watch, dry-run, and delete options.
Source: machine0 changelog.
https://machine0.io/changelog
Blaxel
Website: blaxel.ai
Blaxel is a cloud infrastructure platform for autonomous AI agents, offering persistent sandboxes, agent hosting, and related tooling so agents can run software tasks like a human user would.
Category: Compute & Interaction Environments.
Funding stage: Seed.
Founded: 2024.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Compute & Interaction Environments because it provides persistent digital workspaces and sandboxes where AI agents can operate software, retain state, and execute tasks with low-latency resumes.
Funding rounds
-
seed
announced 2025-12-01
: 7300000 USD
; investors: First Round Capital, Y Combinator, Liquid 2, Transpose, Multimodal
; source: Blaxel Blog
; https://blaxel.ai/blog/Blaxel-Raises-7-3M-Seed-Round-led-by-First-Round-to-Build-Cloud-Infrastructure-for-the-AI-Agent-Eco-23247e47b1ea8067b923d998364e3ced
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/blaxel-ai
-
2026-07-21:
Sapiom wallets can provision Blaxel sandboxes
(product)
. Blaxel and Sapiom announced that agents can use Sapiom wallets to provision perpetual sandboxes, run batch jobs, and deploy applications with spend controls.
Source: Blaxel Blog.
https://blaxel.ai/blog/now-available-blaxel-and-sapiom-team-up-to-let-agents-buy-their-own-sandboxes
-
2026-07-21:
Agent provisioning partnership with Sapiom
(gtm)
. Blaxel announced that AI agents can buy and provision sandboxes, batch jobs, and deployments through Sapiom wallets with spend controls.
Source: Blaxel and Sapiom: AI Agents Can Buy Sandboxes | Blaxel Blog.
https://blaxel.ai/blog/now-available-blaxel-and-sapiom-team-up-to-let-agents-buy-their-own-sandboxes
-
2026-07-09:
Agent Drive surfaced in a public event recap
(gtm)
. Blaxel used a conference recap and product page to promote Agent Drive as a shared filesystem for agents and to explain cross-sandbox state sharing.
Source: Shared filesystems for agents at MCP Dev Summit Mumbai 2026 | Blaxel Blog.
https://blaxel.ai/blog/shared-filesystems-for-agents-at-mcp-dev-summit-mumbai-2026
-
2026-06-25:
Customer proof for enterprise workflow automation
(gtm)
. Blaxel published a Runwork case study showing stateful sandboxes supporting collaborative enterprise AI workflows with 10x faster execution.
Source: Runwork accelerates collaborative enterprise AI workflows by 10x using Blaxel | Blaxel Blog.
https://blaxel.ai/blog/runwork-accelerates-collaborative-enterprise-ai-workflows-by-10x-using-blaxel
-
2026-06-11:
Claude Managed Agents support via self-hosted sandboxes
(product)
. Blaxel added support for Claude Managed Agents as a self-hosted sandbox provider, letting Anthropic manage orchestration while Blaxel runs code execution, files, and network access.
Source: Blaxel Blog.
https://blaxel.ai/blog/blaxel-supports-claude-managed-agents-with-self-hosted-sandbox-execution
E2B
Website: e2b.dev
E2B builds secure cloud sandboxes and virtual machines that give AI agents a persistent environment to run code, use browsers, and complete multi-step tasks like a human operator.
Category: Compute & Interaction Environments.
Funding stage: Series A.
Founded: 2023.
Country: Czech Republic.
Employees: 11-50 employees.
Landscape fit: It fits Compute & Interaction Environments because its core product is a persistent cloud workspace for agents to execute software, browse, and interact with tools rather than just generate text.
Company timeline
-
2026-07-21:
Devin Outposts execution layer launched
(product)
. E2B positioned its microVMs as the execution layer for Devin Outposts, letting sessions run inside customer-controlled infrastructure with BYOC and snapshot support.
Source: E2B Blog.
https://e2b.dev/blog/devin-outposts
-
2026-07-21:
Devin Outposts execution-layer guide
(gtm)
. E2B positioned itself as the execution layer for Devin Outposts, paired with BYOC and snapshot support.
Source: E2B Blog.
https://e2b.dev/blog/devin-outposts
-
2026-07-20:
Anything case study shows consumer-scale adoption
(gtm)
. E2B published a case study showing Anything using E2B to power app generation for more than one million users.
Source: E2B Blog.
https://e2b.dev/blog/anything
-
2026-06-10:
Stripe Projects provisioning for E2B sandboxes
(product)
. E2B added support for provisioning and authenticating sandboxes through Stripe Projects, making sandbox setup agent-driven and credential-scoped.
Source: E2B Blog.
https://e2b.dev/blog/e2b-sandboxes-are-now-available-through-stripe-projects
-
2026-06-10:
Stripe Projects availability
(gtm)
. E2B joined the Stripe Projects developer preview so agents can provision and authenticate sandboxes through Stripe.
Source: E2B Blog.
https://e2b.dev/blog/e2b-sandboxes-are-now-available-through-stripe-projects
-
2026-06-08:
StackAI case study expands enterprise use cases
(gtm)
. E2B showed StackAI using its runtime for custom code, computer use, RPA, and browser use in regulated-industry workflows.
Source: E2B Blog.
https://e2b.dev/blog/stackai
Orgo
Website: orgo.ai
Orgo provides instant cloud desktop computers and sandboxed virtual machines that AI agents can see, control, and use to browse, click, type, run code, and persist files like a human user.
Category: Compute & Interaction Environments.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits this category because its core product is persistent digital workspace infrastructure for agents—a cloud desktop/VM layer that lets agents operate software directly rather than only calling APIs.
Company timeline
-
2025-12-04:
Cloud-desktop vs headless-browser comparison
(product)
. Orgo published a technical comparison arguing that cloud desktops are better for multi-application agent workflows than headless browsers. ([orgo.ai](https://www.orgo.ai/blog/cloud-desktops-vs-headless-browsers))
Source: Orgo blog.
https://www.orgo.ai/blog/cloud-desktops-vs-headless-browsers
-
2025-11-25:
Computer-use positioning expands
(product)
. Orgo published a long-form explainer defining computer use and positioning its desktops as infrastructure for production-scale computer-use agents. ([orgo.ai](https://www.orgo.ai/blog/what-is-computer-use))
Source: Orgo blog.
https://www.orgo.ai/blog/what-is-computer-use
-
2025-06-27:
Open-source agent ecosystem content motion
(gtm)
. Orgo published a setup guide for Agent S2, showing users how to run an open-source computer-use agent locally or on Orgo’s cloud desktop.
Source: Orgo blog.
https://www.orgo.ai/blog/simular-agent-s2-setup-orgo
-
2025-06-27:
Agent S2 cloud-desktop setup guide
(product)
. Orgo published a guide for running the open-source Agent S2 on a local machine or in cloud desktop mode via Orgo. ([orgo.ai](https://www.orgo.ai/blog/simular-agent-s2-setup-orgo?utm_source=openai))
Source: Orgo blog.
https://www.orgo.ai/blog/simular-agent-s2-setup-orgo
-
2025-05-19:
Anthropic Computer Use tutorial established developer onboarding
(gtm)
. Orgo published a hands-on guide showing developers how to set up Anthropic Computer Use with Orgo in 30 seconds, tying the product to the emerging computer-use workflow.
Source: Orgo blog.
https://www.orgo.ai/blog/anthropic-computer-use-setup-orgo
-
2025-05-19:
Anthropic computer-use setup tutorial
(product)
. Orgo published a quickstart for running Anthropic Computer Use on Orgo's virtual desktop, showing direct control methods like clicking, typing, screenshots, and bash commands. ([orgo.ai](https://www.orgo.ai/blog/anthropic-computer-use-setup-orgo?utm_source=openai))
Source: Orgo blog.
https://www.orgo.ai/blog/anthropic-computer-use-setup-orgo
Daytona
Website: daytona.io
Daytona builds secure, elastic sandbox infrastructure for AI agents and developers to run code, use tools, and work in persistent isolated environments.
Category: Compute & Interaction Environments.
Funding stage: Series A.
Founded: 2023.
Country: Croatia.
Employees: 11-50 employees.
Landscape fit: It fits Compute & Interaction Environments because Daytona provides persistent, isolated cloud workspaces/sandboxes that agents can use as computers to operate software, run code, and keep state across sessions.
Funding rounds
-
Seed
announced 2024-06-01
: 5000000 USD
; investors: Upfront Ventures, 500 Emerging Europe
; source: PR Newswire
; https://www.prnewswire.com/news-releases/daytona-secures-5m-to-simplify-development-environments-302181407.html
-
Pre-Seed
announced 2023-11-01
: 2000000 USD
; investors: 500 Emerging Europe, Tiny.vc, Silicon Gardens, Firestreak Ventures
; source: Daytona
; https://www.daytona.io/dotfiles/daytona-raises-2m-in-pre-seed
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 11-50 employees employees.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/daytonaio
-
2026-07-20:
Showed Office automation inside Windows sandboxes
(gtm)
. Daytona published a walkthrough of Microsoft Office running in Windows sandboxes and reusable snapshots for document-automation workflows.
Source: Microsoft Office in Daytona Windows Sandboxes.
https://www.daytona.io/dotfiles/microsoft-office-in-daytona-windows-sandboxes
-
2026-07-07:
Expanded into Windows sandboxes for computer-use workflows
(gtm)
. Daytona launched Windows sandboxes for real Windows environments, positioning the product for desktop app testing, RPA, and Windows-specific automation.
Source: Windows Sandboxes.
https://www.daytona.io/dotfiles/windows-sandboxes
-
2026-04-27:
Docs search, git clone, and API error handling improved
(product)
. Daytona added docs search, OSS deploy guides, optional git CLI clones for large repos, and better HTTP 400 handling for quota and limit errors.
Source: Changelogs · Daytona.
https://www.daytona.io/changelog
-
2026-04-23:
Runtime network controls added
(product)
. Daytona improved sandbox control with runtime network settings, safer declarative builds, Java SDK docs, and Python WebSocket handling.
Source: Changelogs · Daytona.
https://www.daytona.io/changelog
-
2026-04-14:
Sandbox fork and snapshot endpoints shipped
(product)
. Daytona added fork and snapshot API endpoints, along with related dashboard and daemon updates for sandbox lifecycle management.
Source: Changelogs · Daytona.
https://www.daytona.io/changelog
Cua
Website: cua.ai
Cua builds cloud desktop and sandbox infrastructure that lets AI agents run, evaluate, and train on real Linux, Windows, macOS, and Android environments.
Category: Compute & Interaction Environments.
Funding stage: Pre Seed.
Founded: 2025.
Country: United States.
Employees: 3.
Landscape fit: It fits Compute & Interaction Environments because the product provides persistent, OS-level desktops and sandboxes that agents can control like human users, including background GUI actions, multi-OS fleets, and cloud computer-use runtimes.
Company timeline
-
2026-06-29:
Cross-platform input modes and verified actions land
(product)
. Cua Driver extended delivery modes across Linux and macOS, improved verified action reporting, and added more MCP client support.
Source: Cua Changelog.
https://cua.ai/changelog/2026-06-29
-
2026-06-22:
Desktop-scope actions expand on Windows, macOS, and Linux
(product)
. Cua Driver added desktop-scope capture/actions, stronger Wayland coverage, and more explicit error handling across OSes.
Source: Cua Changelog.
https://cua.ai/changelog/2026-06-22
-
2026-06-15:
Driver gains native Wayland preview and diagnostics
(product)
. Cua Driver added a native Wayland preview, driver health reports, Linux ARM64 builds, and macOS cursor/TCC fixes.
Source: Cua Changelog.
https://cua.ai/changelog/2026-06-15
-
2026-06-01:
Driver multi-agent sessions and Python wrapper ship
(product)
. Cua Driver added parallel sessions, a stable macOS signing identity, a PyPI wrapper, and Windows background input support.
Source: Cua Changelog.
https://cua.ai/changelog/2026-06-01
-
2026-03-01:
Cua sponsors ClawCon Madrid
(gtm)
. Following ClawCon SF, Cua extended the event motion internationally by sponsoring ClawCon Madrid.
Source: Cua About.
https://cua.ai/about
-
2026-02-01:
ClawCon SF and CuaBot launch
(gtm)
. Cua hosted ClawCon in San Francisco and used the event to launch CuaBot, described as multi-player computer-use for coding agents.
Source: Cua About.
https://cua.ai/about
Kernel
Website: kernel.sh
Kernel builds browser infrastructure for AI agents, including cloud browsers, managed authentication, observability, and anti-bot handling so agents can use the web like humans do. ([kernel.sh](https://www.kernel.sh/?utm_source=openai))
Category: Compute & Interaction Environments.
Funding stage: Series A.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Compute & Interaction Environments because it provides a persistent cloud browser workspace and related infrastructure that agents use to interact with software and the open web like a human user would. ([kernel.sh](https://www.kernel.sh/?utm_source=openai))
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/trykernel
Steel
Website: steel.dev
Steel builds open-source cloud browser infrastructure that lets AI agents and software automate the web through a persistent, human-like browser workspace.
Category: Compute & Interaction Environments.
Funding stage: Series A.
Founded: 2024.
Country: Canada.
Employees: 2-10 employees.
Landscape fit: It fits Compute & Interaction Environments because it provides agents with persistent cloud browser sessions, credentials, profiles, and observability so they can operate software like a human user would.
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/steeldotdev
-
2026-06-26:
Pricing shifted to metered Launch, Scale, and Enterprise plans
(product)
. Steel closed Launch Week v3 with a simpler pricing model based on browser hours, proxy bandwidth, CAPTCHA solves, and usage credits.
Source: Steel Blog.
https://steel.dev/blog/pricing-update
-
2026-06-26:
Launch Week v3 reset Steel’s packaging around usage-based pricing
(gtm)
. Steel introduced Launch, Scale, and Enterprise plans with metered usage, visible billing, credits for trial, and enterprise features such as SSO and reserved browser pools.
Source: Steel blog.
https://steel.dev/blog/pricing-update
-
2026-06-25:
Atlas launched as a deep research harness
(product)
. Steel introduced Atlas, an open-source deep research harness that turns web evidence into cited reports with coverage and citation grounding.
Source: Steel Launch Week.
https://steel.dev/launch-week
-
2026-06-24:
Native Rust and Go SDKs shipped
(product)
. Steel released typed Rust and Go SDKs, plus cookbook examples, to let teams drive Steel from backend services and agent infrastructure written in those languages.
Source: Steel Blog.
https://steel.dev/blog/rust-go-sdks
-
2026-06-24:
Rust and Go SDKs extended Steel into backend stacks
(gtm)
. Steel released native Rust and Go SDKs to let teams use Steel from the stacks where agent services already run.
Source: Steel blog.
https://steel.dev/blog/rust-go-sdks
Hyperbrowser
Website: hyperbrowser.ai
Hyperbrowser is a cloud browser infrastructure platform that gives AI agents and developers persistent, scalable browser sessions for web automation, scraping, form-filling, and other live-web interactions.
Category: Compute & Interaction Environments.
Founded: 2021.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Compute & Interaction Environments because it provides the persistent digital browser workspace agents use to interact with software and websites like a human user would.
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 11-50 employees employees.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/hyperbrowser
Browserbase
Website: browserbase.com
Browserbase is a cloud browser infrastructure platform that gives AI agents and developers persistent, observable browser environments to browse and interact with the web like humans.
Category: Compute & Interaction Environments.
Funding stage: Series B.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits the category because it provides persistent cloud browser workspaces and automation primitives specifically for AI agents to operate software through a browser, which is a compute-and-interaction environment for agents.
Funding rounds
-
Series B
announced 2025-06-01
: 40000000 USD
; investors: Notable Capital, CRV, Kleiner Perkins
; source: Browserbase
; https://www.browserbase.com/blog/series-b-and-beyond
-
Seed
announced 2024-06-01
: 6500000 USD
; investors: Kleiner Perkins, Basecase Capital, AI Grant, Guillermo Rauch, Andrew Miklas, Barry McCardel
; source: VentureBeat
; https://venturebeat.com/ai/exclusive-browserbase-launches-headless-browser-platform-that-lets-llms-automate-web-tasks
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 11-50 employees employees.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/browserbasehq
-
2026-07-15:
MP4 session recordings became downloadable
(product)
. Browserbase added MP4 session recording downloads through the API and dashboard.
Source: Browserbase Changelog.
https://www.browserbase.com/changelog
-
2026-06-30:
Browserbase Agents launched
(product)
. Browserbase introduced a managed browser-agent product that turns a natural-language task into a runnable agent through a single API call.
Source: Browserbase Changelog.
https://www.browserbase.com/changelog
-
2026-06-30:
Browserbase Agents launched as a managed agent product
(gtm)
. Browserbase introduced Browserbase Agents, a managed browser agent created from natural-language prompts and generally available on every plan. ([browserbase.com](https://www.browserbase.com/blog/introducing-browserbase-agents?utm_source=openai))
Source: Browserbase blog.
https://www.browserbase.com/blog/introducing-browserbase-agents
-
2026-06-11:
Browserbase launched its inaugural Navigate conference
(gtm)
. Browserbase announced Navigate 2026, a one-day conference in San Francisco for teams shipping agents in production, with speakers and attendees from major AI and developer companies. ([browserbase.com](https://www.browserbase.com/blog/navigate?utm_source=openai))
Source: Browserbase blog.
https://www.browserbase.com/blog/navigate
-
2026-05-21:
Browserbase Fetch launched and later expanded to markdown and JSON output
(product)
. Browserbase launched Fetch for reading web pages without a browser session, then updated it to return markdown or structured JSON and raised the data cap.
Source: Browserbase Changelog.
https://www.browserbase.com/changelog
Agent Tool Access & Execution
Infrastructure that gives AI agents direct access to external software and allows them to execute actions through secure machine-readable interfaces.
Composio
Website: composio.dev
Composio builds an agent infrastructure layer that gives AI agents secure, machine-readable access to 1,000+ external apps and lets them authenticate, trigger, and execute actions across those tools.
Category: Agent Tool Access & Execution.
Funding stage: Series A.
Founded: 2023.
Country: unknown.
Employees: 11-50 employees.
Landscape fit: It fits the category because its core product is precisely the middleware that connects AI agents to external software systems, handles delegated auth, and executes actions through APIs and sandboxed workbenches.
Funding rounds
-
Series A
announced 2025-07-01
: 25000000 USD
; investors: Lightspeed Venture Partners, Guillermo Rauch, Dharmesh Shah, Gokul Rajaram, Soham Mazumdar, SV Angel, Blitzscaling Ventures, Operator Partners, Agent Fund by Yohei Nakajima, Elevation Capital, Together Fund
; source: PR Newswire
; https://www.prnewswire.com/news-releases/composio-raises-29m-to-solve-ais-learning-problem-building-skills-that-actually-improve-over-time-302510684.html
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 11-50 employees employees.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/composiohq
-
2026-03-27:
Universal CLI launch
(product)
. Composio released a Universal CLI for direct terminal access to search, execute, link accounts, proxy raw API calls, and run inline scripts across 1,000+ app integrations. ([composio.dev](https://composio.dev/blog/announcing-universal-cli-by-composio?utm_source=openai))
Source: Composio blog.
https://composio.dev/blog/announcing-universal-cli-by-composio
-
2026-03-23:
Custom tools and toolkits expand the platform
(gtm)
. Composio added experimental support for custom tools and toolkits that run locally alongside remote Composio tools within a session.
Source: Composio Docs Changelog.
https://docs.composio.dev/docs/changelog/2026/03/23
-
2025-10-01:
Tool Router beta launches as the core build path
(gtm)
. Composio launched Tool Router beta as the official way to build more complex agent experiences, with a pre-signed MCP-style session flow and more control over toolkits and auth.
Source: Composio Blog.
https://composio.dev/blog/introducing-tool-router-%28beta%29
-
2025-10-01:
Tool Router beta
(product)
. Composio launched Tool Router beta as the official way to build on top of Rube, giving users a pre-signed MCP session and control over supported toolkits and authentication. ([composio.dev](https://composio.dev/blog/introducing-tool-router-%28beta%29?utm_source=openai))
Source: Composio blog.
https://composio.dev/blog/introducing-tool-router-%28beta%29
-
2025-07-01:
New SDK beta and production-ready API posture
(gtm)
. Composio launched a new SDK beta and repositioned the product around more explicit, production-ready flows with clearer auth config and user_id handling.
Source: Composio Blog.
https://composio.dev/blog/new-sdk-preview
Wato
Website: watolabs.com
Wato is a shared AI workspace for companies that gives AI agents permissioned access to memory, tools, workflows, cloud sessions, and artifacts through MCP and other integrations.
Category: Agent Tool Access & Execution.
Funding stage: Seed.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Agent Tool Access & Execution because its core product connects agents to external tools and company systems via MCP, permissions, tracing, and cloud execution so agents can take actions securely inside real workflows.
Company timeline
-
2026-06-01:
Public launch of Wato's shared AI workspace
(product)
. Wato's June 2026 public materials introduced the product as a shared AI workspace built around MCP, versioned team memory, approved connectors, reusable skills/workflows, cloud agent sessions, live artifacts, and traces. ([watolabs.com](https://www.watolabs.com/))
Source: Wato website and blog.
https://www.watolabs.com/
Metorial
Website: metorial.com
Metorial builds agentic infrastructure that lets AI agents securely connect to external tools, data, and workflows through integrations, identity, access control, tracing, and policy enforcement.
Category: Agent Tool Access & Execution.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Agent Tool Access & Execution because its core product gives AI agents machine-readable access to external software and lets them take actions through controlled integrations, SSO, permissions, tracing, and guardrails.
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/metorial
-
2026-06-10:
Portals expands to company-wide access
(gtm)
. Metorial positioned Portals as governed AI access for the whole company and then made it available on Scale, suggesting a move from builder tooling toward organization-wide rollout.
Source: Metorial policies and access control page.
https://metorial.com/features/policies-access-control
-
2026-05-28:
Managed integrations as a central distribution surface
(gtm)
. Metorial formalized its 1000+ managed integrations offering and positioned it as a production-ready catalog available through Portals and the marketplace.
Source: Metorial managed integrations page.
https://metorial.com/features/managed-integrations
-
2026-05-19:
Skills, portals, and agent testing mature the platform
(product)
. Metorial added skill sharing, portal availability on Scale, broader integration access, and Agent Explorer for testing integrations with a real model.
Source: Metorial Blog and Features pages.
https://metorial.com/blog
-
2026-03-24:
EU region and broader enterprise controls ship
(product)
. Metorial added a dedicated EU region along with access controls, files, scale, and hardened auth.
Source: Metorial Changelog.
https://metorial.com/changelog
-
2026-03-24:
EU region and broader enterprise scaling
(gtm)
. Metorial added a dedicated EU region alongside stronger access controls, files in workflows, and scale/security improvements.
Source: Metorial changelog.
https://metorial.com/changelog
Rindler
Website: rindler.ai
Rindler builds an MCP-based infrastructure layer that turns websites into structured, reliable tools so AI agents can log in, retrieve data, and execute actions on sites without APIs.
Category: Agent Tool Access & Execution.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits this category because it gives AI agents secure machine-readable access to external software and websites, with structured execution of actions rather than just data retrieval.
Company timeline
-
2026-07-10:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/rindler
Agent Memory & Context
Infrastructure that preserves knowledge across tasks and sessions so AI agents can maintain continuity and use previously learned context.
Hyper
Website: heyhyper.ai
Hyper is a company-memory layer for AI employees that ingests information from company tools and turns it into a live knowledge graph so agents stay up to date across tasks and sessions.
Category: Agent Memory & Context.
Founded: 2026.
Country: United States.
Landscape fit: It fits Agent Memory & Context because the product is explicitly designed to preserve company knowledge across chats, tools, and sessions so AI agents can remember prior context and act on it consistently.
Company timeline
-
2026-04-13:
Company launch with shared-memory positioning
(gtm)
. Hyper introduced itself as a shared memory layer for AI agents and launched a service for cloud-hosted storage, retrieval, and synchronization of structured knowledge for coding assistants and team members.
Source: Hyper Terms of Service.
https://heyhyper.ai/terms
-
2026-04-13:
Public service launch as a shared memory layer for AI agents
(product)
. Hyper publicly described its service as a shared memory layer for AI agents, with Google OAuth sign-in, workspace-scoped data, third-party integrations, encryption, and a no-training-on-customer-data policy.
Source: Hyper Terms of Service.
https://heyhyper.ai/terms
Memory Store
Website: memory.store
Memory Store builds a shared memory layer for AI agents and teammates, turning Slack, Gmail, meeting notes, and other tool activity into persistent context that agents can reuse across sessions.
Category: Agent Memory & Context.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits the Agent Memory & Context category because its core product is explicitly a cross-session memory layer that preserves decisions, conversations, and context so AI agents do not start each task from scratch.
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/memorystore
-
2026-05-14:
Public product framing as an evolving memory layer
(product)
. Memory Store published an early essay arguing that AI agents are stateless and positioning the company as an intelligent, evolving memory layer for AI.
Source: Memory Store blog.
https://memory.store/blog/why-do-we-need-memory-for-ai
-
2026-04-29:
Content campaign around cross-tool AI memory
(gtm)
. Memory Store published a set of blog posts in late April and early May explaining how to save ChatGPT context and sync AI context across tools automatically, reinforcing its cross-tool memory positioning.
Source: Memory Store blog.
https://memory.store/blog/how-to-save-chatgpt-conversations-stop-repeating-yourself-2026
-
2026-03-10:
Claude and ChatGPT setup guide added
(product)
. Memory Store added a dedicated guide for using the product with Claude and ChatGPT for persistent memory across conversations.
Source: Memory Store guides.
https://memory.store/guides
-
2026-01-08:
Setup guides launch for key AI tools
(gtm)
. Memory Store introduced setup guides for Claude, ChatGPT, Cursor, Raycast, and a general user guide, making onboarding for individual users and developers a first-class GTM surface.
Source: Memory Store guides.
https://memory.store/guides
-
2026-01-08:
Setup guides launched for core AI tools
(product)
. Memory Store published setup guides and a user guide for Claude, ChatGPT, Cursor, Raycast, and prompt templates.
Source: Memory Store guides.
https://memory.store/guides
Hyperspell
Website: hyperspell.com
Hyperspell is a memory layer for AI agents that connects tools like Slack, Gmail, Notion, and Google Drive so agents can recall, remember, and learn over time. ([docs.hyperspell.com](https://docs.hyperspell.com/core/introduction))
Category: Agent Memory & Context.
Funding stage: Pre Seed.
Founded: 2024.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Agent Memory & Context because its core product persists and assembles company knowledge across sessions and tools so agents can use prior context instead of starting from zero. ([docs.hyperspell.com](https://docs.hyperspell.com/core/introduction))
Funding rounds
-
pre-seed
announced 2025-04-01
: 1000000 USD
; investors: Afore Capital, Y Combinator, Scribble Ventures, Pioneer Fund, Alpha Square Group
; source: Hyperspell | LinkedIn
; https://www.linkedin.com/company/hyperspell/
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/hyperspell/
-
2026-07-14:
Query result warnings distinguished outranked sources
(product)
. Hyperspell updated query responses to separate empty-source cases from sources that were present but outranked by others.
Source: Hyperspell Changelog.
https://docs.hyperspell.com/concepts/changelog
-
2026-07-07:
Claude and ChatGPT web connectors became first-class
(product)
. Hyperspell expanded onboarding and MCP docs to support one-click walkthroughs for Claude and ChatGPT web connectors.
Source: Hyperspell Changelog.
https://docs.hyperspell.com/concepts/changelog
-
2026-07-02:
MCP custom connector support launched for Claude
(product)
. Hyperspell added a Claude custom connector that points to its MCP endpoint and authenticates with the Hyperspell dashboard account.
Source: Hyperspell Changelog.
https://docs.hyperspell.com/concepts/changelog
-
2026-06-24:
Failure-modes content for agent memory debugging
(gtm)
. Hyperspell published a practical guide to seven agent memory failure modes and how to diagnose them.
Source: Hyperspell Blog.
https://www.hyperspell.com/blog
-
2026-03-11:
Build-vs-buy content for agent memory infrastructure
(gtm)
. Hyperspell published a technical blog post quantifying the cost of building AI agent memory infrastructure versus buying it.
Source: Hyperspell Blog.
https://www.hyperspell.com/blog
Supermemory
Website: supermemory.ai
Supermemory builds a memory and context infrastructure layer for AI agents and apps, helping systems retain user knowledge, retrieve past context, and personalize responses across sessions.
Category: Agent Memory & Context.
Funding stage: Pre Seed.
Founded: 2024.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Agent Memory & Context because the product is explicitly a persistent memory layer for AI agents, with user profiles, retrieval, and long-term context across tasks and conversations.
Funding rounds
-
seed
announced 2025-10-01
: 2600000 USD
; investors: Susa Ventures, Browder Capital, SF1.vc, Dane Knecht, Jeff Dean, Logan Kilpatrick, David Cramer
; source: TechCrunch
; https://techcrunch.com/2025/10/06/a-19-year-old-nabs-backing-from-google-execs-for-his-ai-memory-startup-supermemory/
-
pre-seed
announced 2025-10-01
: 3000000 USD
; investors: Susa Ventures, Browder Capital, SF1.vc, Dane Knecht, Theo Browne, David Cramer, Julian Weisser
; source: Supermemory
; https://supermemory.ai/blog/supermemory-raises-3-million-and-building-the-best-memory-engine-for-llms/
Company timeline
-
2026-07-20:
Simpler TypeScript SDK search
(product)
. The TypeScript SDK consolidated search into one client.search() call with a searchMode parameter for memories, documents, or hybrid results.
Source: Supermemory API changelog.
https://supermemory.ai/changelog/api/
-
2026-07-04:
Memory review workflow for inferred memories
(product)
. Supermemory added a review queue so users can approve, decline, skip, or undo inferred memories before they become durable context.
Source: Supermemory changelog.
https://supermemory.ai/changelog/api/
-
2026-06-10:
Self-hosted server and MCP governance
(product)
. Supermemory launched a local server for macOS and Linux with encrypted embedded storage and added approval flows for MCP clients before they can access memory.
Source: Supermemory changelog.
https://supermemory.ai/changelog/api/
-
2026-05-28:
SMFS launches as an agent filesystem
(gtm)
. Supermemory launched SMFS, a filesystem-style product for agents, later described as being used by hundreds of companies.
Source: Supermemory blog.
https://supermemory.ai/blog/smfs-making-agentic-retrieval-55-cheaper-and-more-accurate
-
2026-05-18:
Context Cloud positioning expands the platform narrative
(gtm)
. Supermemory reframed its offering as a 'Context Cloud,' bundling memory, retrieval, profiles, reranking, embeddings, and other infrastructure into one stack.
Source: Supermemory blog.
https://supermemory.ai/blog/dear-reader-we-just-made-supermemory-insanely-cheap-the-context-cloud
-
2026-04-27:
SDK v2.0 standardizes the developer integration path
(gtm)
. Supermemory released @supermemory/tools v2.0.0 with a unified API, required conversation identity, and memory saving enabled by default.
Source: Supermemory blog.
https://supermemory.ai/blog/introducing-supermemory-tools-v2-0-0/
Zep AI
Website: getzep.com
Zep AI builds the “context lake” / agent memory infrastructure that lets AI agents remember and reason across chats, documents, events, and business data at scale. ([getzep.com](https://www.getzep.com/about/))
Category: Agent Memory & Context.
Funding stage: Seed.
Founded: 2023.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Agent Memory & Context because its core product is long-term memory/context infrastructure for agents, designed to preserve prior interactions and business context across sessions. ([getzep.com](https://www.getzep.com/about/))
Company timeline
-
2026-07-14:
Emphasized provenance and traceability in agent memory
(gtm)
. Zep published its provenance architecture, showing how retrieved facts can be traced back to raw source data for debugging, compliance, and source-scoped retrieval.
Source: Zep blog.
https://blog.getzep.com/how-zep-tracks-provenance-in-agent-memory/
-
2026-07-09:
ABAC launched for agent memory
(product)
. Zep added attribute-based access control to scope API keys to specific actions and graph data.
Source: Zep Blog.
https://blog.getzep.com/attribute-based-access-control/
-
2026-07-09:
Added ABAC for governed enterprise access
(gtm)
. Zep launched attribute-based access control so customers can attach policies to API keys and restrict endpoints and graph artifacts by policy.
Source: Zep blog.
https://blog.getzep.com/attribute-based-access-control/
-
2026-06-30:
Memory MCP server launched
(product)
. Zep released a Memory MCP Server so desktop assistants and coding tools can share the same governed user graph.
Source: Zep Blog.
https://blog.getzep.com/unified-agent-memory-in-any-mcp-client/
-
2026-06-30:
Launched unified memory across MCP clients
(gtm)
. Zep introduced a Memory MCP Server so desktop assistants, coding tools, and Zep-built agents can share one governed user graph under enterprise SSO.
Source: Zep blog.
https://blog.getzep.com/unified-agent-memory-in-any-mcp-client/
-
2026-06-09:
Batch API launched
(product)
. Zep shipped a new Batch API for loading large backfills, migrations, and document collections into agent memory.
Source: Zep Blog.
https://blog.getzep.com/the-batch-api-load-large-datasets/
Mem0
Website: mem0.ai
Mem0 builds a persistent memory layer for AI agents and applications, letting developers store, retrieve, and update user context across sessions so systems can remember preferences, facts, and prior interactions.
Category: Agent Memory & Context.
Funding stage: Series A.
Founded: 2023.
Country: United States.
Employees: 2-10 employees.
Landscape fit: Mem0 provides persistent memory infrastructure that lets AI agents retain and recall relevant information across interactions.
Funding rounds
-
Series A
announced 2025-10-01
: 20000000 USD
; investors: Basis Set Ventures, Kindred Ventures, Peak XV Partners, GitHub Fund, Y Combinator
; source: Mem0
; https://mem0.ai/series-a
-
Seed
announced 2024-06-01
: 3900000 USD
; investors: Kindred Ventures
; source: Kindred Ventures
; https://kindredventures.com/announcement/mem0-building-the-memory-infrastructure-for-personalized-ai/
Company timeline
-
2026-07-01:
Current careers page shows multiple open roles
(hr)
. Mem0’s careers page currently lists open roles including Backend Engineer, Applied AI Engineer, Full Stack Engineer, and Senior Research Engineer in San Francisco.. Team size observed: 2-10 employees employees.
Source: Mem0 Careers.
https://mem0.ai/careers
-
2026-06-27:
Mem0 adds SDK memory expiration controls
(product)
. Mem0 shipped first-class expiration support across Python and TypeScript SDKs, including expired-memory handling in search and getAll.
Source: Mem0 docs highlights.
https://docs.mem0.ai/changelog/highlights
-
2026-06-10:
Mem0 updates Vercel AI SDK provider to v3
(product)
. Mem0 moved its Vercel AI SDK provider to the v6 provider contract and Mem0 v3 APIs, adding sources in responses and broader deployment controls.
Source: Mem0 docs highlights.
https://docs.mem0.ai/changelog/highlights
-
2026-06-01:
Hiring push for backend and frontend engineers in India
(hr)
. A Mem0 hiring post said the company was seeking two engineers in India: one backend engineer and one frontend engineer.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/paurushmittal_we-are-hiring-at-mem0-2-engineers-in-india-activity-7463214378302144512-RTwH
-
2026-06-01:
Opened founding technical recruiter role
(hr)
. Mem0 posted a LinkedIn job for a Founding Technical Recruiter working directly with the founders and hiring managers.. Team size observed: 2-10 employees employees.
Source: LinkedIn Jobs.
https://www.linkedin.com/jobs/view/founding-technical-recruiter-at-mem0-4413983859
-
2026-05-21:
Mem0 launches agent-first signup flow
(product)
. Mem0 added an agent-first signup path that lets coding agents create an account and get an API key without a human browser flow.
Source: Mem0 blog.
https://mem0.ai/blog/introducing-agentmode-mem0-signup-without-a-human-in-the-loop
Human & Physical-World Access
Services that allow agents to delegate work to people when a task cannot be completed entirely through software.
HireHumans.io
Website: hirehumans.io
A marketplace where AI agents can delegate real-world and verification tasks to human workers, who complete them and get paid instantly in crypto.
Category: Human & Physical-World Access.
Founded: 2026.
Country: Unknown.
Landscape fit: It fits Human & Physical-World Access because it gives AI agents a way to hand off tasks that require a person in the loop or in the physical world, such as calls, photos, and local verification.
Human Pages
Website: humanpages.ai
Human Pages is a marketplace and API that lets AI agents find, hire, and pay real humans to complete real-world tasks like photography, deliveries, research, calls, and inspections.
Category: Human & Physical-World Access.
Founded: 2026.
Country: Singapore.
Employees: 1 employee.
Landscape fit: It fits this category because its core product is explicitly designed for AI agents to delegate work that cannot be completed entirely in software to human workers in the physical world.
Company timeline
-
2026-07-01:
Team size observed
(hr)
. Public source shows a team size of 1 employee employees.. Team size observed: 1 employee employees.
Source: LinkedIn.
https://sg.linkedin.com/company/humanpages-ai
-
2026-02-25:
Public launch of the Human Pages marketplace
(gtm)
. Human Pages announced it was live, described the network as a place where AI agents find and hire real humans, and invited users to create free profiles and join as founding members.
Source: Human Pages blog.
https://humanpages.ai/blog/get-paid-by-ai-agents-humanpages-is-live
-
2026-02-25:
Human Pages went live
(product)
. The company announced that HumanPages.ai was live and introduced it as the network where AI agents find and hire real humans for paid tasks. ([humanpages.ai](https://humanpages.ai/pt/blog/get-paid-by-ai-agents-humanpages-is-live?utm_source=openai))
Source: Human Pages.
https://humanpages.ai/pt/blog/get-paid-by-ai-agents-humanpages-is-live
-
2026-02-11:
Early use-case proof for hiring humans through the API
(gtm)
. The company published content showing AI agents using the Human Pages API to hire social media marketers, signaling an initial applied use case beyond abstract protocol education.
Source: Human Pages blog.
https://humanpages.ai/blog
-
2026-02-08:
Developer-led positioning around MCP-based human hiring
(gtm)
. Human Pages published technical content explaining how AI agents discover and hire people through the Model Context Protocol, framing the product as infrastructure for agents rather than a traditional marketplace.
Source: Human Pages blog.
https://humanpages.ai/blog/mcp-protocol-ai-agents
-
2026-02-08:
Technical product framing around the Human Pages MCP server
(product)
. Human Pages published a technical explainer for its MCP-based hiring flow, describing how agents can search humans by skill and location, retrieve profiles, create job offers, and pay in USDC through the server. ([humanpages.ai](https://humanpages.ai/blog/mcp-protocol-ai-agents?utm_source=openai))
Source: Human Pages.
https://humanpages.ai/blog/mcp-protocol-ai-agents
RentAHuman
Website: rentahuman.ai
Marketplace and API layer that lets AI agents delegate real-world tasks to humans when software alone can’t complete the job.
Category: Human & Physical-World Access.
Funding stage: Seed.
Founded: 2026.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Human & Physical-World Access because the product explicitly connects AI agents with human workers to complete offline, real-world tasks such as deliveries, audits, and fieldwork.
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/rentahuman-ai
-
2026-04-25:
Multi-step orchestration with chained bounties and milestone payments
(product)
. RentAHuman described support for chained bounties, milestone payments, and structured workflows for complex multi-step physical-world tasks.
Source: RentAHuman Blog.
https://rentahuman.ai/blog/multi-step-task-orchestration-complex-workflows
-
2026-04-25:
Programmatic bounties and end-to-end automated hiring
(product)
. RentAHuman documented that agents can post bounties, receive applications, hire humans, communicate, verify delivery, and release payment without manual browser steps.
Source: RentAHuman Blog.
https://rentahuman.ai/blog/programmatic-bounties-post-tasks-without-browser
-
2026-04-25:
Shifted to use-case-led and comparison-based demand generation
(gtm)
. RentAHuman published a broad April content cluster on API-first design, programmatic bounties, escrow, webhooks, global coverage, and comparisons versus legacy gig platforms, which reads like a category-defense and conversion push. ([rentahuman.ai](https://rentahuman.ai/blog/api-first-design-why-rest-api-beats-scraping?utm_source=openai))
Source: RentAHuman Blog.
https://rentahuman.ai/blog/api-first-design-why-rest-api-beats-scraping
-
2026-04-25:
Agent authentication via API keys and MCP tokens
(product)
. RentAHuman documented secure machine-to-machine access using API keys for the REST API and token exchange for MCP, replacing browser-style login assumptions.
Source: RentAHuman Blog.
https://rentahuman.ai/blog/agent-authentication-api-keys-mcp-secure-access
-
2026-03-20:
Expanded into framework-specific developer integrations
(gtm)
. RentAHuman shipped a cluster of tutorials for OpenAI Agents SDK, LangChain, CrewAI, Vercel AI SDK, and AutoGPT, signaling a push to win builders where they already orchestrate agents. ([rentahuman.ai](https://rentahuman.ai/blog?utm_source=openai))
Source: RentAHuman Blog.
https://rentahuman.ai/blog
Humwork
Website: humwork.ai
Humwork is a Y Combinator-backed platform that connects AI agents with verified human experts in real time via API/MCP, so agents can escalate stuck tasks across engineering, design, strategy, finance, legal, and other domains. ([humwork.ai](https://humwork.ai/?utm_source=openai))
Category: Human & Physical-World Access.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Human & Physical-World Access because its core product is a human escalation layer: when an AI agent cannot finish a task in software alone, Humwork routes the task to a real person who solves it and returns the result back into the agent workflow. ([humwork.ai](https://humwork.ai/?utm_source=openai))
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 2-10 employees employees.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/humwork
-
2026-04-08:
Publicly launched the Humwork marketplace/API with credit-based usage and expert onboarding
(gtm)
. Humwork’s legal and privacy pages show the platform operating as a marketplace for real-time consultations between AI agents, professionals, and businesses, with API/MCP access, credits, multiple purchase tiers, expert qualification, and payout flows.
Source: Humwork Terms of Service / Privacy Policy.
https://humwork.ai/terms
Search & Data Access
Infrastructure that gives AI agents access to external information at the moment it is needed for a task.
Crustdata
Website: crustdata.com
Crustdata is a real-time B2B data provider that delivers APIs and webhooks for company and people data, including signals like hires, funding, web traffic, and other live changes. ([crustdata.com](https://www.crustdata.com/?utm_source=openai))
Category: Search & Data Access.
Funding stage: Seed.
Founded: 2023.
Country: United States.
Landscape fit: It fits Search & Data Access because its core product gives AI agents and other workflows on-demand access to fresh external company/person signals at the moment they are needed. ([crustdata.com](https://www.crustdata.com/?utm_source=openai))
Funding rounds
-
Seed
announced 2025-10-01
: 6000000 USD
; investors: Y Combinator, A Capital, General Catalyst, SV Angel, Phosphor Capital
; source: Crustdata
; https://crustdata.com/blog/crustdata-closes-6m-seed-round
Company timeline
-
2026-06-01:
Content-led expansion into recruiting and sourcing workflows
(gtm)
. A June 2026 blog cluster covered lead generation APIs, ATS enrichment, talent mapping, candidate sourcing, and AI-lab talent moves, indicating a broader demand-gen motion around technical recruiting and sourcing systems. ([crustdata.com](https://crustdata.com/blog?utm_source=openai))
Source: Crustdata blog.
https://crustdata.com/blog
-
2026-05-09:
Claude workflows expand Crustdata into multi-step competitive intelligence automation
(product)
. Crustdata published Claude-based workflow guides showing its MCP server powering competitive intelligence, pre-call research, and ABM scoring across enrichment, search, job, post, and web data.
Source: Crustdata blog.
https://crustdata.com/blog/automate-pre-call-account-research-abm-scoring-claude
-
2026-04-09:
MCP connector brings Crustdata into Claude and agent workflows
(product)
. Crustdata introduced an MCP connector through Merge Agent Handler and later documented direct Claude Desktop/Claude Code setups that let agents call company, people, job, posts, and web tools.
Source: Crustdata blog.
https://crustdata.com/blog/agents-crustdata-mcp-connector
-
2026-01-14:
People search API and cached-vs-real-time search model introduced
(product)
. Crustdata published a people search guide describing structured search over professional profiles, with a distinction between cached datasets and real-time APIs for current signals like job changes.
Source: Crustdata blog.
https://crustdata.com/blog/people-search-api
-
2025-12-11:
Data enrichment positioning broadens into a general live enrichment API
(product)
. Crustdata reframed itself as a broader data enrichment layer, highlighting real-time enrichment, webhook-triggered automation, and use cases for lead enrichment, CRM refresh, and trigger-based workflows.
Source: Crustdata blog.
https://crustdata.com/blog/data-enrichment-api
-
2025-10-22:
Seed announcement highlighted customer proof and revenue scale
(gtm)
. In its seed-round announcement, Crustdata said revenue grew from $770K to $4M+ in seven months with zero outbound marketing spend and cited customers including Y Combinator, HubSpot founder Dharmesh Shah via agent.ai, and MNTN. ([crustdata.com](https://crustdata.com/blog/crustdata-closes-6m-seed-round?utm_source=openai))
Source: Crustdata blog.
https://crustdata.com/blog/crustdata-closes-6m-seed-round
Context.dev
Website: context.dev
Context.dev provides a single API that gives AI agents structured, real-time web context from websites, brands, and related online data.
Category: Search & Data Access.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Search & Data Access because the core product is infrastructure that lets agents fetch, structure, and use external web information at task time, rather than relying only on static model knowledge.
Company timeline
-
2026-05-31:
Context.dev backed by Y Combinator
(hr)
. Founder Yahia Bakour announced that Context.dev was now backed by Y Combinator.. Team size observed: 2-10 employees employees.
Source: Context.dev blog.
https://www.context.dev/blog/were-now-backed-by-ycombinator
Firecrawl
Website: firecrawl.dev
Firecrawl provides an API and open-source tooling that turns live websites into clean Markdown and structured data for LLMs, agents, and AI-native software. ([firecrawl.dev](https://www.firecrawl.dev/about?utm_source=openai))
Category: Search & Data Access.
Funding stage: Series A.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Search & Data Access because it gives AI agents real-time access to web data at task time, handling crawling, scraping, and structured output so agents can use current external information. ([firecrawl.dev](https://www.firecrawl.dev/about?utm_source=openai))
Funding rounds
-
Series A
announced 2025-08-01
: 14500000 USD
; investors: Nexus Venture Partners, Y Combinator, Zapier, Tobias Lütke, Abhinav Asthana, Matt McClure
; source: Firecrawl
; https://www.firecrawl.dev/blog/firecrawl-v2-series-a-announcement
Company timeline
-
2026-07-22:
Search relevance model upgrade
(product)
. Firecrawl upgraded /search to return the best excerpts from each result and reported stronger SimpleQA performance with fewer tokens. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
-
2026-06-24:
v2.11 adds research index, keyless access, and privacy controls
(product)
. Firecrawl v2.11 shipped the Research Index, keyless access for core endpoints, automatic PII redaction, deterministicJson, and broader video discovery. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
-
2026-05-26:
Web-scale /monitor launches
(product)
. Firecrawl launched /monitor to watch the web continuously and notify agents when pages change or new matches appear. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
-
2026-04-28:
Fire-PDF and /parse deepen document extraction
(product)
. Firecrawl introduced Fire-PDF and then /parse, moving document handling into a faster Rust-based pipeline for PDFs and office files. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
-
2026-03-25:
/interact launches live browser automation
(product)
. Firecrawl launched /interact to let agents click, type, navigate, and persist browser state in live sessions. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
-
2026-02-03:
v2.8 expands agent workflows and browser tooling
(product)
. Firecrawl v2.8 added parallel agents, the Firecrawl Skill, the Firecrawl CLI, a Spark model family, and agent enhancements such as webhook support and MCP tools. ([firecrawl.dev](https://www.firecrawl.dev/changelog))
Source: Firecrawl Changelog.
https://www.firecrawl.dev/changelog
Parallel Web Systems
Website: parallel.ai
Parallel builds web search, extraction, and research APIs that give AI agents structured, grounded access to live information on the web.
Category: Search & Data Access.
Funding stage: Series B.
Founded: 2023.
Country: United States.
Employees: 51-200 employees.
Landscape fit: It fits the Search & Data Access category because its core products provide agents with real-time external web data, retrieval, and grounding at task time.
Funding rounds
-
series_b
announced 2026-04-01
: 100000000 USD
; investors: Sequoia Capital, Kleiner Perkins, Index Ventures, Khosla Ventures, First Round Capital, Spark Capital, Terrain Capital, Abstract Ventures
; source: PR Newswire
; https://www.prnewswire.com/news-releases/parallel-raises-at-2-billion-valuation-to-scale-web-infrastructure-for-agents-302756350.html
-
series_a
announced 2025-11-01
: 100000000 USD
; investors: Kleiner Perkins, Index Ventures, Spark Capital, Khosla Ventures, First Round Capital, Terrain
; source: Parallel Web Systems
; https://parallel.ai/blog/series-a
Company timeline
-
2026-07-24:
Team size observed
(hr)
. Public source shows a team size of 51-200 employees employees.. Team size observed: 51-200 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/parallel-web/about/
-
2026-07-21:
Responses API launched
(product)
. Parallel added an OpenAI Responses-compatible endpoint for latency-sensitive use cases with fully cited answers, structured output, and 5–60 second responses. ([docs.parallel.ai](https://docs.parallel.ai/resources/changelog))
Source: Parallel changelog.
https://docs.parallel.ai/resources/changelog
-
2026-07-13:
Turbo mode launched for Search
(product)
. Parallel introduced Turbo mode for the Search API, advertising real-time search at $1 per 1,000 requests with p50 latency of 200ms. ([docs.parallel.ai](https://docs.parallel.ai/resources/changelog))
Source: Parallel changelog.
https://docs.parallel.ai/resources/changelog
-
2026-07-10:
Nooks standardized on Parallel for live web grounding
(gtm)
. Parallel published a customer story showing Nooks switched to its Search API and cut web grounding spend while improving production accuracy.
Source: Parallel Web Systems blog.
https://parallel.ai/blog/case-study-nooks
-
2026-06-05:
Launched real-time Entity Search
(gtm)
. Parallel added Entity Search for latency-sensitive company discovery and enrichment workflows.
Source: Parallel Web Systems blog.
https://parallel.ai/blog/entity-search-company
-
2026-05-06:
Monitor API reached general availability
(product)
. Parallel made Monitor generally available with new processors, event streams, snapshots, Basis on every event, domain filtering, and follow-on research interactions. ([docs.parallel.ai](https://docs.parallel.ai/resources/changelog))
Source: Parallel changelog.
https://docs.parallel.ai/resources/changelog
Tavily
Website: tavily.com
AI infrastructure company that provides real-time web search and structured web context so AI agents can access fresh external information at task time. ([tavily.com](https://www.tavily.com/about?utm_source=openai))
Category: Search & Data Access.
Funding stage: Series A.
Founded: 2024.
Country: Israel.
Employees: 51-200 employees.
Landscape fit: It fits Search & Data Access because its core product is an API that lets AI agents search the web, extract live information, and ground responses in fresh context. ([tavily.com](https://www.tavily.com/about?utm_source=openai))
Company timeline
-
2026-07-14:
Keyless search positioning
(gtm)
. Tavily reframed first-use access as keyless search, allowing agents to call Search or Extract without signup, configuration, or an API key.
Source: Tavily Blog.
https://www.tavily.com/blog/What-keyless-search-really-means-for-your-data
-
2026-06-16:
Customer proof from Fieldway
(gtm)
. Tavily published a customer case study showing Fieldway using Tavily to cut a product-research workflow from hours to minutes with cited live sources.
Source: Tavily Blog.
https://www.tavily.com/blog/case-study-fieldway
-
2026-05-27:
x402 no-key access motion
(gtm)
. Tavily announced real-time web search for agents on x402, emphasizing access without an API key.
Source: Tavily Blog.
https://www.tavily.com/blog/tavily-on-x402-real-time-web-search-for-agents-no-api-key-required
-
2026-05-01:
Snowflake and MongoDB ecosystem integrations
(gtm)
. Tavily highlighted availability inside Snowflake Intelligence and hybrid search with MongoDB in its April shipping update.
Source: Tavily Blog.
https://www.tavily.com/blog/whats-new-april
-
2026-05-01:
Research positioning hardens into a building block
(gtm)
. In April, Tavily positioned /research as a reusable building block for modern AI systems and cited use in products and analyst-style workflows.
Source: Tavily Blog.
https://www.tavily.com/blog/whats-new-april
-
2026-04-27:
Arcade.dev integration launch
(gtm)
. Tavily became available in Arcade.dev so agent teams can access cited research through governed execution infrastructure.
Source: Tavily Blog.
https://www.tavily.com/blog/tavily-is-now-available-in-arcade-dev-credible-cited-research-for-ai-agents-to-act-on
Exa
Website: exa.ai
Exa is an AI search engine and search API platform that gives developers and AI agents real-time access to web information, crawling, and deep research tools.
Category: Search & Data Access.
Funding stage: Series C.
Founded: 2021.
Country: United States.
Employees: 51-200 employees.
Landscape fit: It fits Search & Data Access because its core product is retrieval infrastructure that lets AI systems query, crawl, and ground answers in external web data at the moment they need it.
Funding rounds
-
Series C
announced 2026-05-01
: 250000000 USD
; investors: Andreessen Horowitz, Benchmark, Lightspeed, Y Combinator, NVentures, Igor Babuschkin, Scott Wu, Tal Broda
; source: Exa Blog
; https://exa.ai/blog/announcing-series-c
-
Series B
announced 2025-09-01
: 85000000 USD
; investors: Benchmark, Lightspeed, Y Combinator, NVentures
; source: Exa Blog
; https://exa.ai/blog/announcing-series-b
Company timeline
-
2026-07-23:
Launch academic publication search
(product)
. Exa added state-of-the-art search over research papers and technical publications through a publication category.
Source: Exa Blog.
https://exa.ai/blog/publications-search
-
2026-07-23:
Benchmark push expands into academic-publication search
(gtm)
. Exa launched SOTA search over academic publications, extending its benchmark-led positioning into a new research-heavy vertical. ([exa.ai](https://exa.ai/blog?utm_source=openai))
Source: Exa Blog.
https://exa.ai/blog
-
2026-06-16:
Launch Exa Agent
(product)
. Exa released Exa Agent, a single API for frontier web research capabilities at lower cost.
Source: Exa Blog.
https://exa.ai/blog/exa-agent
-
2026-06-16:
Exa Agent launches as a single API for frontier web research
(gtm)
. Exa released Exa Agent to package its retrieval stack into a higher-level API for deep research, list-building, and entity enrichment. ([exa.ai](https://exa.ai/blog/exa-agent?from_theconsensus=1&utm_source=openai))
Source: Exa Blog.
https://exa.ai/blog/exa-agent
-
2026-05-20:
Hiring cluster across sales, solutions, and marketing
(hr)
. Exa's Series C announcement said the company was hiring across sales, solutions, and marketing worldwide.. Team size observed: 51-200 employees employees.
Source: Exa Blog.
https://exa.ai/blog/announcing-series-c
-
2026-05-01:
Series C
(funding)
. Amount: 250000000 USD. Investors: Andreessen Horowitz, Benchmark, Lightspeed, Y Combinator, NVentures, Igor Babuschkin, Scott Wu, Tal Broda
Source: Exa Blog.
https://exa.ai/blog/announcing-series-c
Channel3
Website: trychannel3.com
Channel3 builds an API and product-data infrastructure layer for agentic commerce, organizing product data so AI agents can reliably search, recommend, and purchase products. ([trychannel3.com](https://trychannel3.com/))
Category: Search & Data Access.
Funding stage: Seed.
Founded: 2025.
Country: United States.
Employees: 2-10 employees.
Landscape fit: Channel3 gives AI agents structured product data they can search and use during shopping and transaction workflows.
Funding rounds
-
Seed
announced 2025-12-01
: 6000000 USD
; investors: Matrix, Y Combinator, Ludlow Ventures
; source: Channel3 Blog
; https://trychannel3.com/blog/channel3-raises-6m
-
Pre-Seed
announced 2025-09-01
: 500000 USD
; investors: Y Combinator
; source: Crunchbase
; https://www.crunchbase.com/funding_round/channel3-pre-seed--1c56f0d1
-
Seed
announced 2025-09-01
: 500000 USD
; investors: Y Combinator
; source: Dealroom
; https://app.dealroom.co/companies/channel3
Company timeline
-
2026-05-26:
Developer comparison content for product-data APIs
(gtm)
. Channel3 published a first-person research post comparing product data APIs and arguing that its approach is better suited for AI shopping agents, signaling a content-led developer acquisition motion. ([trychannel3.com](https://trychannel3.com/blog))
Source: Channel3 Blog.
https://trychannel3.com/blog
-
2026-05-12:
Awin Power 100 recognition
(gtm)
. Channel3 announced it was named to Awin’s Power 100, positioning itself as one of the most innovative affiliate partners and using the recognition to reinforce its affiliate-distribution story for advertisers and brands. ([trychannel3.com](https://trychannel3.com/blog/channel3-awin-power-100))
Source: Channel3 Blog.
https://trychannel3.com/blog/channel3-awin-power-100
-
2026-05-01:
Hiring cluster: engineers and Chief of Staff / Head of GTM
(hr)
. Founder posts and the YC jobs page show Channel3 recruiting engineers in NYC and a Chief of Staff / Head of GTM role to support company operations and go-to-market.. Team size observed: 2-10 employees employees.
Source: LinkedIn + Y Combinator jobs.
https://www.linkedin.com/posts/alexanderschiff_channel3-is-hiring-engineers-in-nyc-were-activity-7450996730105499648-yPdG
-
2026-04-01:
Chief of Staff / Head of GTM / Strategy + Ops opening
(hr)
. Channel3 posted a senior in-person New York role covering GTM, partnerships, hiring, and operations.
Source: LinkedIn.
https://www.linkedin.com/jobs/view/chief-of-staff-head-of-gtm-strategy-%2B-ops-at-channel3-4397670273
-
2026-03-11:
Channel3 MCP launched for direct agent access
(product)
. Channel3 added an MCP server that lets AI agents search tens of millions of products in real time, with a free tier that requires no API key and optional paid usage with an API key.
Source: Channel3 Blog / Docs.
https://docs.trychannel3.com/mcp-overview
-
2026-03-01:
Expanded distribution through MCP and agent demo surface
(gtm)
. Channel3 introduced an MCP surface for agents and refreshed demo applications for shopping and storefront use cases, making it easier for AI-native developers to plug into its product graph. ([docs.trychannel3.com](https://docs.trychannel3.com/mcp-overview?utm_source=openai))
Source: Channel3 docs.
https://docs.trychannel3.com/mcp-overview
Redpine
Website: redpine.ai
Redpine is an AI data platform that gives builders and autonomous agents access to licensed, high-quality, multimodal data for training, retrieval, and transactions at scale.
Category: Search & Data Access.
Funding stage: Seed.
Founded: 2024.
Country: Sweden.
Employees: 2-10 employees.
Landscape fit: Redpine gives AI agents direct access to licensed, high-quality data through machine-readable interfaces.
Funding rounds
-
seed
announced 2026-04-01
: 8000000 USD
; investors: NordicNinja, Luminar Ventures, node.vc, Peter Sarlin, Patrik Tran, Anna Nordell Westling
; source: Redpine
; https://www.redpine.ai/insights/redpine-raises-8m-seed
-
pre-seed
announced 2025-09-01
: 1300000 USD
; investors: Colin M. Evans, Feng Hong, Anna Nordell Westling, Daniel Langkilde, Gustav Lindqvist, Spotify alumni (Greens)
; source: Redpine
; https://www.redpine.ai/insights/redpine-to-solve-ais-biggest-bottleneck---launches-with
-
angel / seed
announced 2025-09-01
: 1.3M USD
; investors: Colin M. Evans, Feng Hong, Anna Nordell Westling, Daniel Langkilde, Gustav Lindqvist, Spotify alumni network
; source: Redpine AI
; https://www.redpine.ai/blog/redpine-launch
Company timeline
-
2026-05-13:
Partnership with All Ears
(gtm)
. Redpine launched a partnership with All Ears, making spoken-media signals queryable through Redpine's API and MCP workflows.
Source: Redpine partners with All Ears to bring 50,000 hours of daily spoken media to AI.
https://www.redpine.ai/insights/redpine-partners-with-all-ears
-
2026-05-13:
All Ears partnership adds live spoken media to the platform
(product)
. Redpine announced a live partnership with All Ears, making structured spoken-media data from podcasts, broadcasts, web TV, and social media queryable through Redpine's API and MCP workflows.
Source: Redpine partners with All Ears to bring 50,000 hours of daily spoken media to AI.
https://www.redpine.ai/insights/redpine-partners-with-all-ears
-
2026-05-06:
Brand refresh and new site launch
(gtm)
. Redpine introduced a refreshed identity and new website after its seed round, framing the company as a brand that must speak to both human content owners and AI agents.
Source: Redpine has a new look.
https://www.redpine.ai/insights/redpine-has-a-new-look
-
2026-05-06:
New site and visual identity launched
(product)
. Redpine published a refreshed brand system and new site experience, positioning the company more explicitly around human knowledge, AI agents, and licensed data access.
Source: Redpine has a new look.
https://www.redpine.ai/insights/redpine-has-a-new-look
-
2026-05-01:
Founding GTM US opening
(hr)
. Redpine posted a hiring listing for a founding GTM role to lead its US presence from New York or San Francisco.. Team size observed: 2-10 employees employees.
Source: LinkedIn jobs.
https://www.linkedin.com/jobs/view/gtm-us-at-redpineai-4404113116
-
2026-04-28:
Seed round surfaces customer proof and global expansion
(gtm)
. Redpine announced an $8 million seed round and publicly cited work with leading AI labs, AsedaSciences, and global clients such as Chegg while expanding GTM hiring.
Source: Redpine seed announcement.
https://www.redpine.ai/insights/redpine-raises-8m-seed
Nyne.ai
Website: nyne.ai
Nyne.ai builds an AI-ready people and company data platform that gives agents API access to structured identity, contact, intent, and business context across the web.
Category: Search & Data Access.
Funding stage: Seed.
Founded: 2023.
Country: United States.
Employees: 11-50 employees.
Landscape fit: Nyne.ai gives AI agents direct API access to structured people, identity and company data.
Funding rounds
-
Seed
announced 2026-03-01
: 5300000 USD
; investors: Wischoff Ventures, South Park Commons, Karman Ventures, Gil Elbaz, Soleio, Amr Al-Shihabi, Selin Kocalar, Danh Trang, Evan Tana, Aditya Agarwal
; source: TechCrunch
; https://techcrunch.com/2026/03/13/nyne-founded-by-a-father-son-duo-gives-ai-agents-the-human-context-theyre-missing/
-
Pre-Seed
announced 2025-09-01
: 400000 USD
; investors: South Park Commons
; source: Signal
; https://signal.nfx.com/investors/evan-tana
Company timeline
-
2026-05-01:
Open roles listed on careers page
(hr)
. Nyne.ai's careers page shows multiple open roles across backend engineering, product design, growth marketing, and customer success.. Team size observed: 11-50 employees employees.
Source: Nyne.ai careers page.
https://nyne.ai/careers/
-
2026-04-01:
Careers page lists multiple open roles
(hr)
. Nyne.ai’s careers page listed open roles including Senior Software Engineer (Backend), Product Designer, Growth Marketer, and Customer Success Lead.
Source: Nyne.ai Careers.
https://nyne.ai/careers/
-
2026-04-01:
First SDR started
(hr)
. Nyne.ai announced that Allison Meins started as the company’s first SDR.
Source: LinkedIn.
https://www.linkedin.com/company/nyne-ai
-
2026-04-01:
Founding AE join announced
(hr)
. Michael Fanous announced that Kaleb Castro joined Nyne.ai as Founding AE.
Source: LinkedIn.
https://www.linkedin.com/company/nyne-ai
-
2026-04-01:
CRO hire highlighted
(hr)
. Michael Fanous publicly spotlighted Marc Zaks as CRO at Nyne.ai, describing his background across revenue and customer leadership roles.
Source: LinkedIn.
https://www.linkedin.com/company/nyne-ai
-
2026-03-01:
Seed
(funding)
. Amount: 5300000 USD. Investors: Wischoff Ventures, South Park Commons, Karman Ventures, Gil Elbaz, Soleio, Amr Al-Shihabi, Selin Kocalar, Danh Trang, Evan Tana, Aditya Agarwal
Source: TechCrunch.
https://techcrunch.com/2026/03/13/nyne-founded-by-a-father-son-duo-gives-ai-agents-the-human-context-theyre-missing/
