AI Context & Data Infrastructure contains 25 companies across
3 categories.
Infrastructure that turns proprietary data into usable, retrievable and persistent context for AI applications and agents.
AI Databases & Retrieval
Systems that store, index and retrieve information for AI applications. This includes vector databases, multimodel databases and retrieval engines built around semantic or hybrid search.
Barrel DB
Website: barrel-db.eu
Barrel DB builds open-source, AI-native databases in Erlang, including a vector database for semantic/hybrid search and a document database for replication and sync.
Category: AI Databases & Retrieval.
Founded: 2025.
Country: France.
Landscape fit: It fits AI Databases & Retrieval because its core products are a vector database with hybrid search and an AI-oriented retrieval stack, plus a document database designed for AI workloads and embeddings. ([barrel-db.eu](https://barrel-db.eu/?utm_source=openai))
Company timeline
-
2026-06-14:
Barrel Vector 2.0.0 removed cluster, HTTP, and gateway features
(product)
. Barrel Vector 2.0.0 removed the Ra/Raft cluster mesh, sharding, scatter-gather, the HTTP server, and the multi-tenant gateway, leaving the product embedded-only. ([github.com](https://github.com/barrel-db/barrel_vectordb/blob/main/CHANGELOG.md))
Source: barrel_vectordb changelog.
https://github.com/barrel-db/barrel_vectordb/blob/main/CHANGELOG.md
-
2026-06-01:
Built a technical proof stack around quick starts, benchmarks, and embedding service
(gtm)
. The docs introduced quick-start flows, performance benchmarks, and a separate embedding library/service to support the core database products.
Source: Barrel DB docs.
https://docs.barrel-db.eu/
-
2026-06-01:
Introduced managed hosting as an upsell path
(gtm)
. The homepage added a direct invitation to contact the company for managed hosting alongside the self-hosted open-source offer.
Source: Barrel DB homepage.
https://barrel-db.eu/
-
2026-06-01:
Expanded the product story to a two-database platform
(gtm)
. The site and docs moved from a single-database story to a combined platform narrative: Barrel Vector for semantic/hybrid search and Barrel Docs for replication and sync.
Source: Barrel DB homepage and docs.
https://barrel-db.eu/
-
2026-06-01:
Public launch of an AI-native open-source database platform
(gtm)
. Barrel DB’s public website positioned the company as an open-source, embeddable, production-ready platform for vector and document databases built in Erlang.
Source: Barrel DB homepage.
https://barrel-db.eu/
-
2026-04-04:
Barrel DocDB release introduced views and OpenAPI
(product)
. Barrel DocDB v0.5.0 added map/reduce views, an OpenAPI 3.0 specification with Swagger UI, and a richer fold_docs API, while also migrating sequence handling to HLC timestamps. ([github.com](https://github.com/barrel-db/barrel_docdb/releases))
Source: barrel_docdb releases.
https://github.com/barrel-db/barrel_docdb/releases
Qdrant
Website: qdrant.tech
Qdrant is an open-source vector database and cloud retrieval platform for semantic search, RAG, recommendation systems, and other AI applications. ([qdrant.tech](https://qdrant.tech/about-us/))
Category: AI Databases & Retrieval.
Funding stage: Series B.
Founded: 2021.
Country: Germany.
Employees: 100+ employees.
Landscape fit: It fits AI Databases & Retrieval because Qdrant builds infrastructure for storing, indexing, and retrieving high-dimensional data for semantic/hybrid search and RAG workflows. ([qdrant.tech](https://qdrant.tech/about-us/))
Funding rounds
-
Series B
announced 2026-03-01
: 50000000 USD
; investors: AVP, Bosch Ventures, Unusual Ventures, Spark Capital, 42CAP
; source: Qdrant
; https://qdrant.tech/blog/series-b-announcement/
-
Series A
announced 2024-01-01
: 28000000 USD
; investors: Spark Capital, Unusual Ventures, 42CAP
; source: Qdrant
; https://qdrant.tech/blog/series-a-funding-round/
-
Seed
announced 2023-04-01
: 7500000 USD
; investors: Unusual Ventures, 42CAP, IBB Ventures, Amr Awadallah, Junior Kim
; source: Qdrant
; https://qdrant.tech/articles/seed-round/
-
Pre-Seed
announced 2022-11-01
: 2000000 EUR
; investors: 42CAP, IBB Ventures, Business angels
; source: IBB Ventures
; https://www.ibbventures.de/en/news/qdrant-financing
Company timeline
-
2026-05-11:
Qdrant 1.18 shipped TurboQuant and new operational controls
(product)
. Qdrant 1.18 added TurboQuant, memory monitoring, add/remove named vectors, audit log querying, request tracing IDs, per-collection metrics, and stricter guardrails. ([qdrant.tech](https://qdrant.tech/blog/qdrant-1.18.x/))
Source: Qdrant Blog.
https://qdrant.tech/blog/qdrant-1.18.x/
-
2026-04-28:
Qdrant Cloud enterprise launch added GPU indexing, Multi-AZ, and audit logging
(product)
. Qdrant Cloud added GPU-accelerated indexing, Multi-AZ support, and audit logging for higher-throughput and higher-availability enterprise workloads. ([qdrant.tech](https://qdrant.tech/blog/qdrant-cloud-enterprise-launch/))
Source: Qdrant Blog.
https://qdrant.tech/blog/qdrant-cloud-enterprise-launch/
-
2026-04-28:
Qdrant Cloud adds GPU indexing, Multi-AZ, and audit logging
(gtm)
. Qdrant Cloud introduced infrastructure features aimed at higher-throughput, higher-availability, and compliance-sensitive workloads.
Source: Qdrant blog.
https://qdrant.tech/blog/qdrant-cloud-enterprise-launch/
-
2026-03-31:
Qdrant Skills for AI Agents launched
(product)
. Qdrant released Skills as an agent-facing knowledge layer that encodes Qdrant operational and retrieval guidance for AI agents and humans. ([qdrant.tech](https://qdrant.tech/blog/qdrant-skills-release/))
Source: Qdrant Blog.
https://qdrant.tech/blog/qdrant-skills-release/
-
2026-03-01:
Series B
(funding)
. Amount: 50000000 USD. Investors: AVP, Bosch Ventures, Unusual Ventures, Spark Capital, 42CAP
Source: Qdrant.
https://qdrant.tech/blog/series-b-announcement/
-
2026-02-20:
Qdrant 1.17 added relevance feedback and stronger observability
(product)
. Qdrant 1.17 introduced relevance feedback queries, lower-latency search controls, cluster-wide telemetry, segment optimization monitoring, and audit logging. ([qdrant.tech](https://qdrant.tech/blog/qdrant-1.17.x/))
Source: Qdrant Blog.
https://qdrant.tech/blog/qdrant-1.17.x/
Weaviate
Website: weaviate.io
Weaviate builds an AI-native, open-source vector database and retrieval engine for storing, indexing, and semantically searching unstructured data for AI applications.
Category: AI Databases & Retrieval.
Funding stage: Series B.
Founded: 2019.
Country: Netherlands.
Employees: 90+ employees.
Landscape fit: It fits this category because its core product is a vector database/retrieval system designed for semantic search, RAG, and other AI workloads over unstructured data.
Funding rounds
-
Series B
announced 2023-04-01
: 50000000 USD
; investors: Index Ventures, Battery Ventures, New Enterprise Associates, Cortical Ventures, Zetta Venture Partners, ING Ventures
; source: PR Newswire
; https://www.prnewswire.com/news-releases/weaviate-raises-50-million-series-b-funding-to-meet-soaring-demand-for-ai-native-vector-database-technology-301803274.html
-
Series A
announced 2022-02-01
: 16000000 USD
; investors: New Enterprise Associates, Cortical Ventures, Zetta Venture Partners, ING Ventures
; source: PR Newswire
; https://www.prnewswire.com/news-releases/semi-technologies-16m-series-a-round-highlights-a-new-wave-of-ai-first-database-tech-301486766.html
-
Seed
announced 2020-08-01
: 1600000 USD
; investors: Zetta Venture Partners, ING Ventures
; source: PR Newswire
; https://www.prnewswire.com/news-releases/semi-technologies-16m-series-a-round-highlights-a-new-wave-of-ai-first-database-tech-301486766.html
Company timeline
-
2026-06-25:
HFresh and MCP Server reached GA
(product)
. Weaviate 1.38 moved HFresh and the built-in MCP Server to general availability and reworked cluster-wide async replication to run from a single scheduler.
Source: Weaviate 1.38 Release.
https://weaviate.io/blog/weaviate-1-38-release
-
2026-06-25:
Weaviate 1.38 continued platform hardening and agent tooling
(gtm)
. Weaviate 1.38 shipped the HFresh disk-based vector index and made the MCP Server generally available, continuing the move toward production agent infrastructure.
Source: Weaviate blog.
https://weaviate.io/blog/weaviate-1-38-release
-
2026-06-17:
Weaviate Cloud became free to start
(product)
. Weaviate Cloud added free tiers across the database, Query Agent, and Engram, removing the credit-card and expiration requirement.
Source: Weaviate Cloud is now free to start.
https://weaviate.io/blog/weaviate-free-tier
-
2026-06-17:
Weaviate Cloud became free to start across the platform
(gtm)
. Weaviate removed the initial barrier to trying its database, Query Agent, and Engram by making Cloud free to start across the product suite.
Source: Weaviate blog.
https://weaviate.io/blog/weaviate-free-tier
-
2026-06-03:
Engram launched as a managed memory service for agents
(gtm)
. Weaviate made Engram generally available in Cloud as a managed memory and context service for agentic applications.
Source: Weaviate blog.
https://weaviate.io/blog/engram-generally-available
-
2026-05-28:
RBAC expanded for Cloud organizations
(gtm)
. Weaviate added Editor and Viewer roles to Cloud console RBAC, making access control more granular for teams and enterprises.
Source: Weaviate blog.
https://weaviate.io/blog/rbac-overview
SurrealDB
Website: surrealdb.com
SurrealDB is a multi-model, AI-native database platform that combines relational, document, graph, time-series, vector, geospatial, and key-value data with built-in search and retrieval for modern applications.
Category: AI Databases & Retrieval.
Founded: 2021.
Country: United Kingdom.
Landscape fit: It fits AI Databases & Retrieval because SurrealDB explicitly supports vector, full-text, and hybrid search plus graph/context-aware querying in a single database built for AI-native applications.
Funding rounds
-
Series A extension
announced 2026-02-01
: 23000000 USD
; investors: Chalfen Ventures, Begin Capital, FirstMark, Georgian
; source: SurrealDB
; https://surrealdb.com/blog/surrealdb-raises-23m-series-a-extension-to-power-the-ai-native-database-era
-
Series A
announced 2024-06-01
: 20000000 USD
; investors: FirstMark, Georgian, Crew Capital, Alumni Ventures
; source: SurrealDB
; https://surrealdb.com/blog/surrealdb-raises-20m-to-disrupt-database-tech-introduces-new-cloud-beta-access
-
Seed
announced 2023-01-01
: 6000000 USD
; investors: FirstMark Capital, Matt Turck
; source: SurrealDB
; https://surrealdb.com/blog/we-are-thrilled-to-announce-our-6m-seed-round-led-by-firstmark-capital-and-matt-turck
Company timeline
-
2026-07-02:
SurrealDB 3.2 opened a new graph-query surface
(product)
. SurrealDB 3.2 introduced experimental ISO GQL/OpenGQL and exposed GraphQL/OpenGQL over RPC and the first-party MCP server.
Source: SurrealDB releases.
https://surrealdb.com/releases/3.2
-
2026-07-02:
Launch of Scale tier in SurrealDB Cloud
(gtm)
. SurrealDB launched Scale, a new cloud tier aimed at high availability, fault tolerance, and production traffic for real applications and AI agents. ([surrealdb.com](https://surrealdb.com/blog/introducing-scale-surrealdb-cloud-built-for-high-availability-and-scale?utm_source=openai))
Source: SurrealDB Blog.
https://surrealdb.com/blog/introducing-scale-surrealdb-cloud-built-for-high-availability-and-scale
-
2026-06-30:
SurrealDB Cloud Scale launched
(product)
. SurrealDB introduced Scale, a new cloud tier built for high-availability, multi-node, multi-zone production workloads.
Source: SurrealDB blog.
https://surrealdb.com/blog/introducing-scale-surrealdb-cloud-built-for-high-availability-and-scale
-
2026-06-09:
Nebius AI Cloud Marketplace launch
(gtm)
. SurrealDB became available on the Nebius AI Cloud Marketplace, extending its cloud distribution and aligning the product with an AI infrastructure channel. ([surrealdb.com](https://surrealdb.com/blog/surrealdb-is-now-available-on-nebius-ai-cloud-marketplace))
Source: SurrealDB Blog.
https://surrealdb.com/blog/surrealdb-is-now-available-on-the-nebius-ai-cloud-marketplace
-
2026-05-27:
SurrealDB 3.1 added DiskANN, MCP support, and enterprise ops tooling
(product)
. SurrealDB 3.1 introduced DiskANN, a built-in MCP server, audit logging, slow-query tooling, and a new security-oriented release process.
Source: SurrealDB blog.
https://surrealdb.com/blog/surrealdb-3-1-stability-diskann-and-a-new-release-process
-
2026-02-17:
SurrealDB 3.0 launched with new AI-agent positioning
(product)
. SurrealDB 3.0 shipped as a stable release and repositioned the product around AI agent memory, improved developer experience, and major engine foundations.
Source: SurrealDB releases.
https://surrealdb.com/releases/3.0
Vespa.ai
Website: vespa.ai
Vespa.ai builds Vespa, an AI search platform and distributed serving engine for retrieval, ranking, machine learning inference, and real-time serving over large-scale text, vector, tensor, and structured data.
Category: AI Databases & Retrieval.
Funding stage: Series A.
Founded: 2023.
Country: Norway.
Employees: 51-200 employees.
Landscape fit: It fits AI Databases & Retrieval because its core product is a system for indexing, retrieving, ranking, and serving data for AI applications, including semantic/vector and hybrid search use cases.
Funding rounds
-
Series A
announced 2023-11-01
: 31000000 USD
; investors: Blossom Capital
; source: Vespa.ai
; https://vespa.ai/2023-11-01-blossom-funding/
Company timeline
-
2026-07-01:
Careers pages show active recruiting and 2027 internship timing
(hr)
. Vespa.ai's careers page says the company is always hiring, and the students page lists software developer recruiting plus Summer Internship 2027 applications opening September 1, 2026.. Team size observed: 51-200 employees employees.
Source: Vespa.ai careers and students pages.
https://vespa.ai/careers/
-
2026-05-29:
Agentic retrieval proof with Metal AI
(gtm)
. Vespa published a customer story on Metal AI, highlighting an agent-driven intelligence platform where most retrieval is handled by AI agents.
Source: Vespa Blog.
https://blog.vespa.ai/how-metal-ai-built-an-agent-driven-intelligence-platform-on-vespa-cloud/
-
2026-02-16:
Self-serve onboarding and marketplace distribution
(gtm)
. Vespa added a playground, a Kubernetes operator, browser-based query/data feeding, and GCP Marketplace availability to improve adoption and procurement.
Source: Vespa Blog.
https://blog.vespa.ai/vespa-newsletter-february-2026/
-
2025-12-03:
Automated ANN tuning and faster vector-distance computation
(product)
. Vespa added an ANN autotune tool in PyVespa and accelerated exact vector distance using Google Highway, alongside new monitoring metrics for ANN behavior. ([blog.vespa.ai](https://blog.vespa.ai/vespa-newsletter-december-2025/))
Source: Vespa Newsletter, December 2025.
https://blog.vespa.ai/vespa-newsletter-december-2025/
-
2025-09-22:
Digital commerce positioning expands
(gtm)
. Vespa sharpened its commerce message around product discovery, merchandising, personalization, and agentic shopping in a single platform.
Source: Vespa Blog.
https://blog.vespa.ai/powering-the-next-era-of-personalised-commerce/
-
2025-09-12:
ANN tuning, geo filtering, global relevance score, and automatic instance migration
(product)
. Vespa expanded ANN tuning controls, added geoBoundingBox filtering, made global relevanceScore available in final ranking, and automated Vespa Cloud instance migration. ([blog.vespa.ai](https://blog.vespa.ai/vespa-newsletter-september-2025/))
Source: Vespa Newsletter, September 2025 / Vespa Now 2025 year in review.
https://blog.vespa.ai/vespa-newsletter-september-2025/
Meilisearch
Website: meilisearch.com
Meilisearch is a flexible, user-focused search engine and AI retrieval platform for adding fast full-text, hybrid, and semantic search to websites and applications.
Category: AI Databases & Retrieval.
Funding stage: Series A.
Founded: 2019.
Country: France.
Employees: 11-50 employees.
Landscape fit: It fits AI Databases & Retrieval because it provides the search backend and retrieval layer for applications, including full-text search, semantic/hybrid search, and indexing/retrieval infrastructure for AI-enabled products.
Funding rounds
-
Series A
announced 2022-10-01
: 15000000 USD
; investors: Felicis, Guillermo Rauch, CRV, Mango Capital, LocalGlobe, Seedcamp
; source: Meilisearch
; https://www.meilisearch.com/blog/meilisearch-series-a
-
Seed
announced 2022-01-01
: 5000000 USD
; investors: CRV, Robin Vasan, Mango Capital, Guillermo Rauch, LocalGlobe, Seedcamp
; source: Meilisearch
; https://www.meilisearch.com/blog/meilisearch-raised-5meu-seed-fundraising
-
Seed
announced 2020-08-01
: 1500000 EUR
; investors: LocalGlobe, Seedcamp, Kima, Tiny.vc
; source: Meilisearch
; https://www.meilisearch.com/blog/meili-fundraise
Company timeline
-
2026-06-04:
CarJager story shows migration-led customer proof for Cloud
(gtm)
. In June 2026, Meilisearch highlighted CarJager’s migration from Algolia to Meilisearch Cloud, emphasizing lower costs, no DevOps overhead, and million-search scale. ([meilisearch.com](https://www.meilisearch.com/blog))
Source: How CarJager runs millions of vehicle searches every month on Meilisearch, with zero DevOps.
https://www.meilisearch.com/blog/how-carjager-runs-millions-of-vehicle-searches-every-month-on-meilisearch-with-zero-devops
-
2026-04-20:
Q1 2026 launch week adds horizontal scaling and enterprise security
(gtm)
. In April 2026, Meilisearch shipped sharding and replication, then SSO/SAML plus MFA, and wrapped the week with chat UI and other launch-week releases aimed at production-scale buyers. ([meilisearch.com](https://www.meilisearch.com/blog/launch-week-q1-2026-wrap-up?utm_source=openai))
Source: Launch Week wrap-up: everything that shipped in five days, April 2026.
https://www.meilisearch.com/blog/launch-week-q1-2026-wrap-up
-
2026-04-15:
Meilisearch Cloud added a built-in Chat UI
(product)
. Cloud added a chat interface with auto-generated prompts, guardrails, an inspector tab, and integration snippets. ([meilisearch.com](https://www.meilisearch.com/blog/chat-ui))
Source: Meilisearch blog.
https://www.meilisearch.com/blog/chat-ui
-
2026-04-13:
Meilisearch Cloud launched sharding and replication
(product)
. Meilisearch Cloud added managed sharding and replication for horizontal scale, availability, and lower-latency serving. ([meilisearch.com](https://www.meilisearch.com/blog/sharding-replication))
Source: Meilisearch blog.
https://www.meilisearch.com/blog/sharding-replication
-
2026-02-01:
Cloud enterprise page consolidates scale, security, and sales-led proof
(gtm)
. By 2026, Meilisearch’s enterprise landing page centered the sales motion on security, compliance, observability, sharding, SLAs, and proof points like Hugging Face’s scale use case. ([meilisearch.com](https://www.meilisearch.com/solutions/enterprise))
Source: Enterprise search | Meilisearch Solutions.
https://www.meilisearch.com/solutions/enterprise
-
2025-11-01:
Marketing hiring push for two roles
(hr)
. Maya Shin said she was looking for two people to join the marketing team, including a Growth Marketer.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/mayyashin_meilisearch-careers-activity-7396102321266393088-3bxI
Pinecone
Website: pinecone.io
Pinecone is a vector database and knowledge engine for building AI applications that need fast semantic retrieval, search, and long-term memory.
Category: AI Databases & Retrieval.
Funding stage: Series B.
Founded: 2019.
Country: United States.
Employees: 137 employees.
Landscape fit: It fits AI Databases & Retrieval because its core product stores embeddings and supports semantic search, hybrid retrieval, and knowledge access for AI applications.
Funding rounds
-
Series B
announced 2023-04-01
: 100000000 USD
; investors: Andreessen Horowitz, ICONIQ Growth, Menlo Ventures, Wing Venture Capital
; source: Pinecone
; https://www.pinecone.io/newsroom/pinecone-raises-usd100m-in-series-b-funding-to-provide-long-term-memory-for-ai/
-
Series A
announced 2022-03-01
: 28000000 USD
; investors: Menlo Ventures, Tiger Global, Wing Venture Capital
; source: Pinecone
; https://www.pinecone.io/newsroom/pinecone-announces-usd28m-series-a-financing-to-bring-search-into-ai-age/
-
Seed
announced 2021-01-01
: 10000000 USD
; investors: Wing Venture Capital
; source: Pinecone
; https://www.pinecone.io/blog/announcing-vector-database/
Company timeline
-
2026-07-01:
Pinecone Nexus entered public preview
(product)
. Pinecone opened Nexus to public preview as a knowledge engine for agents, with connectors, workspaces, contexts, manifests, sandboxes, and KnowQL. ([pinecone.io](https://www.pinecone.io/blog/pinecone-nexus-public-preview/))
Source: Pinecone Blog.
https://www.pinecone.io/blog/pinecone-nexus-public-preview/
-
2026-07-01:
Nexus public preview
(gtm)
. Pinecone moved Nexus into public preview after early-access testing, with the launch positioned around faster, cheaper agentic retrieval and knowledge compilation. ([pinecone.io](https://www.pinecone.io/newsroom/))
Source: Pinecone Newsroom / Blog.
https://www.pinecone.io/newsroom/
-
2026-07-01:
Pinecone showed an active hiring cluster across engineering and product
(hr)
. Pinecone’s careers page and LinkedIn jobs page currently list multiple open roles, including Senior/Staff Software Engineer positions for Database, Experience, and Search & Retrieval Infrastructure, plus Product and Marketing roles on the careers page.. Team size observed: 137 employees employees.
Source: Pinecone Careers / LinkedIn.
https://www.pinecone.io/careers/?id=5701970003
-
2026-07-01:
Pinecone publicly welcomed Jackson Gold as Associate Field Engineer
(hr)
. Pinecone posted that Jackson Gold joined as a new Associate Field Engineer and said he brings experience in AI and startups.. Team size observed: 137 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/pinecone-io
-
2026-06-03:
Microsoft OneLake integration for Nexus
(gtm)
. Pinecone Nexus integrated with Microsoft OneLake so enterprise AI agents can work directly from Microsoft ecosystem data. ([pinecone.io](https://www.pinecone.io/newsroom/microsoft-onelake-nexus/))
Source: Pinecone Newsroom.
https://www.pinecone.io/newsroom/microsoft-onelake-nexus/
-
2026-05-05:
Pinecone Marketplace entered public preview
(product)
. Pinecone Marketplace launched in public preview as a managed way to build, publish, and operate AI-powered knowledge applications with templates, connectors, evaluation, versioning, and an end-user chat interface. ([docs.pinecone.io](https://docs.pinecone.io/release-notes/2026))
Source: Pinecone Docs / Newsroom.
https://www.pinecone.io/blog/pinecone-nexus-public-preview/
Chroma
Website: trychroma.com
Chroma is an open-source embedding/vector database and retrieval platform for AI applications, with support for storing embeddings, dense and sparse search, metadata filtering, and multimodal retrieval.
Category: AI Databases & Retrieval.
Founded: 2022.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits AI Databases & Retrieval because its core product stores embeddings and provides semantic, hybrid, and metadata-based retrieval for AI apps.
Funding rounds
-
seed
announced 2023-04-01
: 18000000 USD
; investors: Quiet Capital, Naval Ravikant, Max Altman, Jack Altman, Jordan Tigani, Guillermo Rauch, Akshay Kothari, Amjad Masad, Spencer Kimball
; source: Chroma
; https://www.trychroma.com/blog/seed
-
pre-seed
announced 2022-05-01
: 2300000 USD
; investors: AIX Ventures, Bloomberg Beta, AI Grant
; source: Chroma
; https://www.trychroma.com/blog/seed
Company timeline
-
2026-07-01:
Current careers page shows six open roles across sales, engineering, and design
(hr)
. Chroma's careers page currently lists one founding Account Executive, four engineering roles, and one product designer role.. Team size observed: 11-50 employees employees.
Source: Chroma Careers.
https://www.trychroma.com/careers
-
2026-05-01:
Founder said Chroma had grown to 17 people and was hiring operations support
(hr)
. Jeff Huber posted that Chroma had grown to 17 people, that the team was all engineers, and that he was hiring a head of operations in San Francisco.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/jeffchuber_chroma-has-grown-to-17-people-all-engineers-activity-7356113750476681216-G4JQ
-
2026-04-09:
EU region support added
(product)
. Chroma Cloud added an EU deployment region in GCP europe-west1 alongside the existing US region.
Source: Chroma Changelog.
https://www.trychroma.com/changelog/eu-region-support
-
2026-04-09:
Enterprise networking and compliance features mature the cloud offer
(gtm)
. Chroma added private networking for AWS and GCP, and later EU region support for data residency.
Source: Chroma Changelog.
https://www.trychroma.com/changelog/eu-region-support
-
2026-03-04:
Chroma Sync expanded to S3, GitHub, and web
(product)
. Chroma broadened Sync into a serverless ingestion system for S3 buckets, GitHub repositories, and websites.
Source: Chroma Changelog.
https://www.trychroma.com/changelog/chroma-sync-s3-github-and-web
-
2026-02-12:
Metadata arrays launched
(product)
. Chroma added native support for arrays of strings, numbers, and booleans in metadata fields.
Source: Chroma Changelog.
https://www.trychroma.com/changelog/metadata-arrays
turbopuffer
Website: turbopuffer.com
turbopuffer is an object-storage-native search database that combines vector and full-text retrieval for AI applications, optimized for scalable, low-cost semantic search over very large document corpora. ([turbopuffer.com](https://turbopuffer.com/about))
Category: AI Databases & Retrieval.
Funding stage: Seed.
Founded: 2023.
Country: Canada.
Employees: 11-50 employees.
Landscape fit: It fits AI Databases & Retrieval because the product is explicitly a search database for vector and full-text retrieval, built to index and retrieve large-scale data for AI workloads. ([turbopuffer.com](https://turbopuffer.com/about))
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/turbopuffer
-
2026-06-01:
Lower-friction launch pricing and cost-optimized model added
(gtm)
. turbopuffer reduced the Launch plan minimum from $64/month to $16/month and added i8 vector type support for quantization-aware models. ([turbopuffer.com](https://turbopuffer.com/docs/pricing-log))
Source: Pricing Changelog.
https://turbopuffer.com/docs/pricing-log
-
2026-06-01:
Lower entry pricing for Launch plan
(product)
. turbopuffer reduced the Launch plan minimum from $64/month to $16/month and kept usage-based billing above that floor.
Source: Pricing Changelog.
https://turbopuffer.com/docs/pricing-log
-
2026-04-01:
Branching, sparse vectors, and namespace pinning
(product)
. turbopuffer added instant namespace branching, sparse vector search, namespace pinning for cached high-QPS workloads, and asynchronous copy/recall operations.
Source: Roadmap & Changelog.
https://turbopuffer.com/docs/roadmap
-
2026-03-01:
Enterprise packaging hardened with audit logs and BYOC
(gtm)
. By spring 2026, turbopuffer’s product and docs exposed enterprise controls including audit logs, SSO, HIPAA-ready BAA, CMEK, private networking, and BYOC operation models. ([turbopuffer.com](https://turbopuffer.com/pricing?doc=2&docs=10000000&namespace=3&namespaces=1000&utm_source=openai))
Source: turbopuffer pricing and BYOC/security docs.
https://turbopuffer.com/pricing?doc=7&docs=1000000000&namespace=0&namespaces=1&write=0&writes=0
-
2026-02-03:
FTS v2 launched for full-text search
(gtm)
. turbopuffer released FTS v2, a major full-text search upgrade with up to 20x better performance and new relevance features such as rank-by-filter and regex filtering. ([turbopuffer.com](https://turbopuffer.com/blog/fts-v2?utm_source=openai))
Source: FTS v2: up to 20x faster full-text search.
https://turbopuffer.com/blog/fts-v2
LanceDB
Website: lancedb.com
LanceDB is an open-source, AI-native multimodal lakehouse and vector database platform for storing, retrieving, and working with embeddings, images, video, and other AI data.
Category: AI Databases & Retrieval.
Funding stage: Series A.
Founded: 2022.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits AI Databases & Retrieval because it provides vector search, hybrid retrieval, and multimodal data storage/indexing for AI applications, including RAG, semantic search, and training data workflows.
Funding rounds
-
Series A
announced 2025-06-01
: 30000000 USD
; investors: Theory Ventures, CRV, Y Combinator, Databricks Ventures, RunwayML, Zero Prime, Swift
; source: LanceDB
; https://www.lancedb.com/blog/series-a-funding
-
Seed
announced 2024-05-01
: 8000000 USD
; investors: CRV, Y Combinator, Essence VC, Swift Ventures
; source: TechCrunch
; https://techcrunch.com/2024/05/15/lancedb-which-counts-midjourney-as-a-customer-is-building-databases-for-multimodal-ai/
Company timeline
-
2026-06-24:
Feature engineering and blob handling expanded for multimodal training workflows
(product)
. LanceDB added scalable feature engineering guidance for multimodal datasets and later introduced Lance Blob V2 late materialization in Spark plus materialized model features for faster VLM fine-tuning.
Source: LanceDB Blog.
https://www.lancedb.com/blog/faster-vlm-fine-tuning-with-materialized-model-features-in-lancedb
-
2026-06-01:
Senior Product Manager opening visible on LinkedIn
(hr)
. A LinkedIn job posting showed LanceDB hiring for a Senior Product Manager role in San Francisco.. Team size observed: 11-50 employees employees.
Source: LinkedIn Jobs.
https://www.linkedin.com/jobs/view/product-manager-at-lancedb-4412161637
-
2026-05-29:
Reproducible multimodal data curation
(gtm)
. LanceDB positioned curation as an end-to-end multimodal lakehouse workflow, emphasizing reproducibility, versioning, and dataset preparation for training rather than just search. ([lancedb.com](https://www.lancedb.com/blog/reproducible-data-curation-in-the-multimodal-lakehouse?utm_source=openai))
Source: LanceDB blog.
https://www.lancedb.com/blog/reproducible-data-curation-in-the-multimodal-lakehouse
-
2026-04-29:
Enterprise-scale vector search at 10B vectors
(gtm)
. LanceDB highlighted a distributed enterprise architecture for vector search at 10B scale, with segmented indexing and parallel query execution to support larger production deployments. ([lancedb.com](https://www.lancedb.com/blog/how-lancedb-accelerates-vector-search-at-10-billion-scale?utm_source=openai))
Source: LanceDB blog.
https://www.lancedb.com/blog/how-lancedb-accelerates-vector-search-at-10-billion-scale
-
2026-04-02:
DuckDB Lance extension became a core extension
(product)
. LanceDB collaborated with the DuckDB team to make Lance a core DuckDB extension, simplifying installation and making Lance-based SQL retrieval part of the standard DuckDB experience.
Source: LanceDB Blog.
https://www.lancedb.com/blog/agentic-coding-as-community-stewardship
-
2026-03-04:
Native Hugging Face Hub support, branching, and geospatial indexing added
(product)
. LanceDB added native Hugging Face Hub support for publishing and scanning multimodal datasets, introduced a Git-like branching model with shallow clones, and brought geospatial support with GeoArrow and R-Tree indexing.
Source: LanceDB Blog.
https://www.lancedb.com/blog/newsletter-february-2026
Zilliz
Website: zilliz.com
Zilliz is the company behind Milvus, an open-source high-performance vector database and managed cloud platform for AI applications that need semantic search, retrieval, and similarity matching over unstructured data. ([milvus.io](https://milvus.io/intro?utm_source=openai))
Category: AI Databases & Retrieval.
Funding stage: Series B.
Founded: 2017.
Country: China.
Employees: 51-200 employees.
Landscape fit: It fits AI Databases & Retrieval because Milvus is purpose-built for storing, indexing, and retrieving vector embeddings for semantic search, RAG, and multimodal AI workloads. ([milvus.io](https://milvus.io/intro?utm_source=openai))
Funding rounds
-
Series B extension
announced 2022-08-01
: 60000000 USD
; investors: Prosperity7 Ventures, Temasek's Pavilion Capital, Hillhouse Capital, 5Y Capital, Yunqi Capital
; source: Vector Database Company Zilliz Raises $60 Million Series B Extension to Expand Its Operations in Silicon Valley
; https://www.businesswire.com/news/home/20220824005057/en/Vector-Database-Company-Zilliz-Raises-%2460-Million-Series-B-Extension-to-Expand-Its-Operations-in-Silicon-Valley
-
Series B
announced 2020-11-01
: 43000000 USD
; investors: Hillhouse Capital, TrustBridge Capital, Pavilion Capital, 5Y Capital, Yunqi Partners
; source: Zilliz Raises $43M to Build AI-powered Unstructured Data Processing and Analysis Platforms
; https://www.globenewswire.com/news-release/2020/11/17/2127910/0/en/Zilliz-Raises-43M-to-Build-AI-powered-Unstructured-Data-Processing-and-Analysis-Platforms.html
-
Series A
announced 2018-01-01
: 10000000 USD
; investors: Morningside Venture Capital, Green Pine Capital Partners, Yunqi Partners, Eminence Ventures
; source: ZILLIZ Speeds up GPU Database Commercialization Layout After Completing $10 Million A1 Round Financing
; https://www.prnewswire.com/news-releases/zilliz-speeds-up-gpu-database-commercialization-layout-after-completing-10-million-a1-round-financing-300660628.html
Company timeline
-
2026-07-06:
BYOC-I expands to GCP
(gtm)
. Zilliz Cloud Bring Your Own Cloud Infrastructure became available on Google Cloud Platform, extending the enterprise deployment option to another major cloud.
Source: Zilliz Cloud Changelog.
https://docs.zilliz.com/docs/changelogs
-
2026-07-06:
BYOC-I adds multi-dataplane support on GCP
(product)
. Zilliz Cloud BYOC added multiple dataplanes within a single project, allowing a project to span multiple regions with region-specific infrastructure units.
Source: Zilliz Cloud Developer Hub.
https://docs.zilliz.com/docs/changelogs
-
2026-07-01:
Public employee count visible on LinkedIn
(hr)
. Zilliz’s LinkedIn company page showed a company size of 51-200 employees and 147 visible employees.. Team size observed: 51-200 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/zilliz
-
2026-07-01:
Current international sales hiring visible
(hr)
. Zilliz had a live Enterprise Account Executive opening for South Korea on its Lever careers page.. Team size observed: 51-200 employees employees.
Source: Lever.
https://jobs.lever.co/zilliz/a72bb831-5bd7-4ac4-b158-100be65185c5
-
2026-07-01:
Current solutions hiring visible
(hr)
. Zilliz had a live Solutions Architect opening for the SF Bay Area on its Lever careers page.. Team size observed: 51-200 employees employees.
Source: Lever.
https://jobs.lever.co/zilliz/3789b92a-72ee-4d73-9e1a-ec09ba966ed7
-
2026-07-01:
Current sales hiring visible for SF Bay Area
(hr)
. Zilliz had a live Enterprise Account Executive opening for the SF Bay Area on its Lever careers page.. Team size observed: 51-200 employees employees.
Source: Lever.
https://jobs.lever.co/zilliz/3cf825a1-a659-4686-b992-3bf03f5b082c/apply
Memory & Context Infrastructure
Infrastructure that helps AI agents retain, update and reuse information across interactions. These systems manage persistent memory and assemble relevant context when an agent needs it.
Cognee
Website: cognee.ai
Cognee builds an open-source AI memory platform and data layer that helps agents retain, update, and retrieve context across sessions using documents, relational data, and knowledge graphs.
Category: Memory & Context Infrastructure.
Funding stage: Seed.
Founded: 2024.
Country: Germany.
Employees: 11-50 employees.
Landscape fit: It fits Memory & Context Infrastructure because the product is explicitly designed to give AI agents persistent memory, context retrieval, and knowledge updates across interactions, which is the core function of this category.
Funding rounds
-
seed
announced 2026-02-01
: 7500000 USD
; investors: Pebblebed, 42CAP, Vermilion Cliffs Ventures, Google DeepMind angels, n8n angels, Snowplow angels
; source: Cognee
; https://www.cognee.ai/blog/cognee-news/cognee-raises-seven-million-five-hundred-thousand-dollars-seed
-
pre-seed
announced 2024-11-01
: 1500000 USD
; investors: 42CAP, Angel Invest, Combination VC, Bob van Luijt, Alexander Dean
; source: Nordic9
; https://nordic9.com/news/cognee-raises-15-million-pre-seed-with-42cap-angel-invest-combination-vc-and-angel-investors-in-germany
Company timeline
-
2026-07-12:
Topic Index and smarter dataset search
(product)
. Cognee 1.3.0 added an optional Topic Index for dataset clustering, along with batch ingestion improvements and better search ranking.
Source: GitHub Releases.
https://github.com/topoteretes/cognee/releases
-
2026-07-01:
Active hiring page lists three open roles
(hr)
. Cognee's careers page currently shows open roles for Principal Platform Engineer, Principal Python Engineer, and AI Engineer – Working Student, with offices in Germany and the USA and remote options.. Team size observed: 11-50 employees employees.
Source: Cognee Careers.
https://www.cognee.ai/careers
-
2026-06-26:
cognee 1.0 launches as a full platform
(gtm)
. Cognee released 1.0 with a memory-native API, managed cloud, self-hosting, Rust and TypeScript options, and first-party support for Claude Code, Cursor, Codex, OpenClaw, Windsurf, Gemini CLI, and MCP-compatible agents. ([cognee.ai](https://www.cognee.ai/blog/cognee-news/cognee-1-0-announcement?utm_source=openai))
Source: Cognee blog.
https://www.cognee.ai/blog/cognee-news/cognee-1-0-announcement
-
2026-06-26:
cognee 1.0 API repositioning
(product)
. Cognee formally rebuilt its surface around remember, recall, improve, and forget, with session-aware memory and a unified recall flow replacing the older search-centric model.
Source: Cognee blog.
https://www.cognee.ai/blog/cognee-news/cognee-1-0-announcement
-
2026-06-26:
Truth subspace retrieval and learned feedback
(product)
. Cognee 1.2.2 introduced a truth subspace index for distilled session learnings, opt-in feedback weighting for retrieval, and reliability fixes for S3-backed LanceDB setups.
Source: GitHub Releases.
https://github.com/topoteretes/cognee/releases
-
2026-06-21:
Session distillation and new proposal/skill APIs
(product)
. Cognee 1.2.0 added smart session distillation, a proposals API, inline skill ingestion, and stricter security defaults for public registration and backend access.
Source: GitHub Releases.
https://github.com/topoteretes/cognee/releases
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: Memory & Context Infrastructure.
Funding stage: Series A.
Founded: 2023.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits the Memory & Context Infrastructure category because its core product persists and recalls context across interactions for AI agents, replacing ad hoc prompt stuffing and RAG workarounds with a reusable memory layer.
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
Zep
Website: getzep.com
Zep builds persistent memory and context infrastructure for AI agents, using temporal context graphs to retain, update, and retrieve information across chats, documents, events, and business data.
Category: Memory & Context Infrastructure.
Funding stage: Seed.
Founded: 2023.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Memory & Context Infrastructure because its core product is a governed, temporal memory layer that helps agents remember past interactions and assemble relevant context across sessions.
Funding rounds
-
Convertible Note
announced 2024-01-01
: 500000 USD
; investors: Y Combinator
; source: CB Insights
; https://www.cbinsights.com/company/zep-2/financials
Company timeline
-
2026-07-09:
Added ABAC for API keys
(product)
. Zep added attribute-based access control to scope API keys to specific endpoints and graph data they can read. ([blog.getzep.com](https://blog.getzep.com/attribute-based-access-control/))
Source: Zep blog.
https://blog.getzep.com/attribute-based-access-control/
-
2026-07-01:
Careers page shows senior engineering/research hiring
(hr)
. Zep's careers page lists open roles for Applied Research Engineer and Senior AI Engineer, indicating an active hiring push on the core technical team.. Team size observed: 11-50 employees employees.
Source: Zep careers page.
https://www.getzep.com/careers/
-
2026-06-30:
Launched Memory MCP Server
(product)
. Zep introduced a Memory MCP Server so desktop and coding agents can connect to the same governed user graph through enterprise SSO. ([blog.getzep.com](https://blog.getzep.com/unified-agent-memory-in-any-mcp-client/))
Source: Zep blog.
https://blog.getzep.com/unified-agent-memory-in-any-mcp-client/
-
2026-06-30:
Unified memory via MCP and enterprise SSO expands distribution beyond Zep-native apps
(gtm)
. Zep launched a Memory MCP Server that lets Claude, ChatGPT, Cursor, and other MCP clients share governed memory through enterprise identity.
Source: Zep Blog.
https://blog.getzep.com/unified-agent-memory-in-any-mcp-client/
-
2026-06-04:
Rolled out Smart Context Assembly
(product)
. Zep replaced the fixed Context Block assembly method with Smart Context Assembly, which ranks multiple context types together and returns a smaller block with higher accuracy. ([blog.getzep.com](https://blog.getzep.com/smart-context-assembly-fewer-tokens-higher-quality/))
Source: Zep blog.
https://blog.getzep.com/smart-context-assembly-fewer-tokens-higher-quality/
-
2026-05-21:
Introduced Observations as a new context type
(product)
. Zep added Observations, a derived context type that automatically captures recurring patterns, decisions, and preferences across the graph. ([blog.getzep.com](https://blog.getzep.com/observations-durable-patterns-from-the-context-graph/))
Source: Zep blog.
https://blog.getzep.com/observations-durable-patterns-from-the-context-graph/
Letta
Website: letta.com
Letta builds infrastructure for stateful AI agents, focused on persistent memory, context management, and agent runtimes that help models learn from experience over time.
Category: Memory & Context Infrastructure.
Funding stage: Seed.
Founded: 2024.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits Memory & Context Infrastructure because its core product is software that stores, updates, and assembles long-term context for AI agents, including memory blocks, context repositories, and agent runtime tooling.
Funding rounds
-
Seed
announced 2024-09-01
: 10000000 USD
; investors: Felicis, Sunflower Capital, Essence VC
; source: PR Newswire
; https://www.prnewswire.com/news-releases/berkeley-ai-research-lab-spinout-letta-raises-10m-seed-financing-led-by-felicis-to-build-ai-with-memory-302257004.html
Company timeline
-
2026-07-01:
Current careers page shows multiple open research and engineering roles
(hr)
. Letta's Join Us page currently lists open positions for Research Engineer / Scientist in Memory, Post-Training, and Self-Improvement, plus Software Engineer, Agent Harness.. Team size observed: 11-50 employees employees.
Source: Letta careers page.
https://www.letta.com/join-us/
-
2026-06-25:
Mods extend the Letta Code harness
(product)
. Letta introduced Mods as an agent-friendly way to extend and adapt the Letta Code harness.
Source: Letta blog.
https://www.letta.com/blog/towards-agents-that-learn/
-
2026-06-01:
Harness extensions via Mods
(gtm)
. Letta introduced Mods, a catalog and install flow for harness modifications that let agents adapt through code. ([letta.com](https://www.letta.com/agent/mods?utm_source=openai))
Source: Letta.
https://www.letta.com/agent/mods
-
2026-04-06:
Letta Code app launch
(gtm)
. Letta launched the Letta Code app for locally run, deeply personalized agents that keep memory across models and can be extended with skills. ([letta.com](https://www.letta.com/blog/introducing-the-letta-code-app/))
Source: Letta.
https://www.letta.com/blog/introducing-the-letta-code-app/
-
2026-04-06:
Letta Code app launches for local personalized agents
(product)
. Letta released the Letta Code desktop app for Mac, Windows, and Linux, with agent initialization, memory refinement, model portability, and skills support.
Source: Letta blog.
https://www.letta.com/blog/introducing-the-letta-code-app/
-
2026-04-02:
Context Constitution codifies experiential AI principles
(product)
. Letta published the Context Constitution, a set of principles for how agents should manage context to learn from experience.
Source: Letta blog.
https://www.letta.com/blog/context-constitution/
Graphlit
Website: graphlit.com
Graphlit is a developer-focused AI context layer that ingests multimodal data, extracts entities/relationships, and provides search and retrieval APIs so AI agents can use persistent, time-aware organizational memory.
Category: Memory & Context Infrastructure.
Funding stage: Seed.
Founded: 2021.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits the Memory & Context Infrastructure category because Graphlit explicitly positions itself as a context layer/agent memory platform with continuous ingestion, entity resolution, temporal memory, provenance, and context assembly for AI agents. ([graphlit.com](https://www.graphlit.com/?utm_source=openai))
Funding rounds
-
Seed
announced 2025-01-01
: 3560000 USD
; investors: 8VC, Bossa Invest, Feld Ventures, MAVA Ventures, Vectura Ventures, F4 Fund
; source: Dealroom
; https://app.dealroom.co/companies/graphlit
Company timeline
-
2026-07-01:
Closed new signups and began winding down Graphlit
(gtm)
. Graphlit’s pricing page now states that the service is winding down and that new signups are closed while existing accounts remain available during transition.
Source: Graphlit Pricing.
https://www.graphlit.com/pricing
-
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: Graphlit LinkedIn company page.
https://www.linkedin.com/company/graphlit?utm_source=openai
-
2026-02-17:
Zine launched as a personal AI chat product
(product)
. Graphlit introduced Zine, a personal AI chat product grounded in notes, email, files, and connected tools.
Source: Graphlit Blog.
https://www.graphlit.com/blog/introducing-the-new-zine
-
2026-01-19:
Dossium launched as a customer context platform
(product)
. Graphlit introduced Dossium as a team-oriented customer context product built on the Graphlit infrastructure.
Source: Graphlit Blog.
https://www.graphlit.com/blog/introducing-customer-context-platform
-
2026-01-19:
Launched Dossium as a customer context product
(gtm)
. Graphlit announced Dossium as a live customer context platform for sales, support, and customer-success teams, built on Graphlit infrastructure and centered on CRM-linked organizational memory.
Source: Graphlit Blog.
https://www.graphlit.com/blog/introducing-customer-context-platform
-
2026-01-01:
Facts mode in Graphlit Studio
(product)
. Graphlit said it was shipping Facts mode in Studio, with table and graph views for browsing, filtering, and searching extracted facts.
Source: Building the Event Clock.
https://www.graphlit.com/blog/building-the-event-clock
Data Ingestion & Transformation
Infrastructure that converts proprietary documents and live enterprise data into AI-ready context. It handles parsing, extraction, transformation and continuous synchronization before the data reaches retrieval or memory systems.
Pathway
Website: pathway.com
Pathway builds a live data framework for powering AI applications, RAG, ETL, and real-time data processing from continuously changing enterprise data sources.
Category: Data Ingestion & Transformation.
Funding stage: Seed.
Founded: 2020.
Country: France.
Employees: 11-50 employees.
Landscape fit: It fits this category because its core product converts documents and live enterprise data into AI-ready context through ingestion, transformation, and continuous synchronization before retrieval or memory layers.
Funding rounds
-
seed
announced 2024-12-01
: 10000000 USD
; investors: TQ Ventures, Kadmos Capital, Inovo VC, Market One Capital, id4 Ventures, Łukasz Kaiser
; source: Sifted
; https://sifted.eu/articles/pathway-10m-seed-round-news
-
pre-seed
announced 2022-12-01
: 4500000 USD
; investors: Inovo VC, Market One Capital, Roger Crook, Łukasz Kaiser
; source: Sifted
; https://sifted.eu/articles/female-led-deeptech-pathway-ai
Company timeline
-
2026-03-23:
Brand marketing internship added
(hr)
. Pathway’s careers page added a Brand Marketing & PR Internship in Palo Alto.. Team size observed: 11-50 employees employees.
Source: Pathway Careers.
https://pathway.com/careers
-
2026-03-13:
AI benchmark and datasets hiring cluster
(hr)
. Pathway added AI Benchmark & Datasets Engineer / Researcher roles in Palo Alto, Paris, and Wrocław.. Team size observed: 11-50 employees employees.
Source: Pathway Careers.
https://pathway.com/careers
-
2026-02-16:
Autoscaling and snapshotting strengthened production deployment
(product)
. Pathway added worker autoscaling based on workload, MongoDB snapshot output mode, and TLS support for Postgres writes. ([pathway.com](https://pathway.com/developers/user-guide/development/changelog/))
Source: Pathway changelog.
([pathway.com](https://pathway.com/developers/user-guide/development/changelog/))
-
2025-11-14:
Machine learning research hiring cluster
(hr)
. Pathway’s careers page added multiple Machine Learning Researcher / Engineer (Foundational Models) roles in Palo Alto, Paris, and Wrocław.. Team size observed: 11-50 employees employees.
Source: Pathway Careers.
https://pathway.com/careers
-
2025-11-07:
Engineering hiring cluster for Machine Learning DevOps
(hr)
. Pathway added multiple Machine Learning DevOps - Cloud and Compute Cluster - R&D Support roles across Europe and Palo Alto.. Team size observed: 11-50 employees employees.
Source: Pathway Careers.
https://pathway.com/careers
-
2025-08-22:
MCP server launch for live agents
(gtm)
. Pathway launched its MCP Server for live indexing and analytics, extending the product into agent tool distribution and current-state retrieval.
Source: Pathway blog.
https://pathway.com/blog/live-ai-mcp-server
Cradl AI
Website: cradl.ai
Cradl AI is a no-code AI document automation platform that extracts, validates, and structures data from PDFs, scans, images, and forms for downstream workflows and integrations. ([cradl.ai](https://www.cradl.ai/?utm_source=openai))
Category: Data Ingestion & Transformation.
Founded: 2016.
Country: Norway.
Employees: 2-10 employees.
Landscape fit: It fits data ingestion & transformation because the product ingests unstructured documents, extracts fields into structured outputs, and synchronizes that data into tools like n8n, Zapier, Power Automate, email, and webhooks. ([cradl.ai](https://www.cradl.ai/integrations?utm_source=openai))
Funding rounds
-
seed
announced 2018-06-01
: 5000000 NOK
; investors: Finstart Nordic
; source: Nordic9
; https://nordic9.com/news/lucidtech-adds-610k-from-finstart-nordic-news6400818406/
Company timeline
-
2026-07-06:
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/cradl-ai/
-
2026-06-09:
Second-opinion validation launched
(gtm)
. On Jun. 9, 2026, Cradl AI introduced second-opinion validation, using a second model to cross-check predictions before data moves downstream. ([cradl.ai](https://www.cradl.ai/changelog))
Source: Cradl AI Changelog.
https://www.cradl.ai/changelog
-
2026-06-09:
Second opinion validation for downstream review
(product)
. Cradl AI added second opinion validation, which compares predictions against a second model to flag uncertain cases for review and let confident predictions pass automatically.
Source: Cradl AI Changelog.
https://www.cradl.ai/changelog
-
2026-04-01:
Invoice workflow tutorials extended to n8n and Power Automate
(gtm)
. Between February and April 2026, Cradl AI published step-by-step tutorials for extracting invoice data in Power Automate and n8n, turning its core use case into repeatable self-serve acquisition content. ([cradl.ai](https://www.cradl.ai/posts/how-to-extract-data-from-invoices-in-power-automate?utm_source=openai))
Source: Cradl AI Blog.
https://www.cradl.ai/posts/how-to-extract-data-from-invoices-in-power-automate
-
2026-03-19:
Native n8n node launched
(gtm)
. On Mar. 19, 2026, Cradl AI shipped a native n8n node so users can drop document extraction into n8n workflows with built-in human-in-the-loop review, training, and validation. ([cradl.ai](https://www.cradl.ai/changelog))
Source: Cradl AI Changelog.
https://www.cradl.ai/changelog
-
2026-03-19:
Native Cradl AI node for n8n
(product)
. Cradl AI shipped a native n8n node that bundles extraction, human-in-the-loop review, validation, and model training into n8n workflows.
Source: Cradl AI Changelog.
https://www.cradl.ai/changelog
Anyformat
Website: anyformat.ai
Anyformat is a document intelligence platform that parses, extracts, validates, and routes data from documents, images, scans, and audio into structured outputs for enterprise workflows.
Category: Data Ingestion & Transformation.
Funding stage: Seed.
Founded: 2024.
Country: Spain.
Employees: 11-50 employees.
Landscape fit: It fits AI Context & Data Infrastructure because it converts proprietary, unstructured enterprise documents into structured, AI-ready data before that data is sent into downstream systems like ERPs and workflows. The company’s own materials emphasize parsing, extraction, validation, and integration into systems such as SAP, Oracle, and NetSuite, which is exactly the ingestion/transformation layer in this category. ([anyformat.ai](https://www.anyformat.ai/?utm_source=openai))
Funding rounds
-
Seed
announced 2025-11-01
: 3300000 EUR
; investors: Kibo Ventures, 4Founders, Abac Nest Ventures, Decelera Ventures
; source: PR Newswire
; https://www.prnewswire.com/news-releases/anyformat-closes-a-3-3-million-seed-round-led-by-kibo-ventures-with-the-aim-of-transforming-document-management-for-global-corporations-302627516.html
-
Pre-Seed
announced 2024-10-01
: 520000 EUR
; investors: 4Founders Capital, Abac Nest Ventures, business angels
; source: PR Newswire UK
; https://www.prnewswire.co.uk/news-releases/anyformat-raises-520-000-to-revolutionize-the-automation-of-unstructured-data-extraction-and-analysis-302268795.html
Company timeline
-
2026-07-07:
Production benchmark messaging expanded
(product)
. anyformat published benchmark results on 1,000+ real documents and emphasized visual citations plus calibrated confidence as production differentiators. ([anyformat.ai](https://anyformat.ai/blog/anyformat-document-ai-benchmarks))
Source: anyformat blog.
https://anyformat.ai/blog/anyformat-document-ai-benchmarks
-
2026-07-07:
Production benchmark campaign published
(gtm)
. anyformat published benchmark content claiming top performance across multiple real-document studies and emphasizing calibrated confidence and visual citations.
Source: anyformat blog.
https://anyformat.ai/blog/anyformat-document-ai-benchmarks
-
2026-07-07:
API v3 and packet/run model launched
(gtm)
. anyformat released API v3 with document packets and runs as first-class resources, plus cleaner upload and rerun flows.
Source: anyformat changelog.
https://docs.anyformat.ai/changelog/overview
-
2026-06-30:
AI doc assistant Annie launched
(product)
. anyformat introduced Annie, an AI doc assistant inside Studio that sets up workflows from plain-language instructions, sample documents, and existing fields. ([anyformat.ai](https://anyformat.ai/blog/meet-annie-ai-doc-assistant))
Source: anyformat blog.
https://anyformat.ai/blog/meet-annie-ai-doc-assistant
-
2026-06-30:
AI doc assistant Annie launched
(gtm)
. anyformat introduced Annie to set up workflows from plain-language prompts and sample documents, reducing setup from hours to minutes.
Source: anyformat blog.
https://anyformat.ai/blog/meet-annie-ai-doc-assistant
-
2026-06-22:
Business subscription and self-serve pricing introduced
(gtm)
. anyformat launched a self-serve Business plan with recurring monthly credits, alongside Enterprise deployment options.
Source: anyformat changelog.
https://docs.anyformat.ai/changelog/overview
Unstructured
Website: unstructured.io
Unstructured builds an AI data processing platform that converts complex, unstructured enterprise documents and other multimodal sources into clean, structured, AI-ready outputs for retrieval, search, and downstream LLM applications.
Category: Data Ingestion & Transformation.
Funding stage: Series A.
Founded: 2022.
Country: United States.
Employees: 51-200 employees.
Landscape fit: It fits AI Context & Data Infrastructure because its core product ingests proprietary documents, extracts structure, transforms content into AI-friendly formats, and continuously feeds downstream retrieval or memory systems.
Funding rounds
-
Series B
announced 2024-03-01
: 40000000 USD
; investors: Menlo Ventures, Databricks Ventures, IBM Ventures, NVentures, Vivek Ranadivé, Chet Kapoor, Allison Pickens, Madrona, Bain Capital Ventures, Mango Capital
; source: Business Wire
; https://www.businesswire.com/news/home/20240314620374/en/Unstructured-Raises-%2440M-Series-B-From-Menlo-Ventures-Databricks-Ventures-IBM-Ventures-and-NVIDIA-to-Make-Enterprise-Data-LLM-ready
-
Seed and Series A
announced 2023-07-01
: 25000000 USD
; investors: Bain Capital Ventures, Madrona, M12 Ventures, Mango Capital, MongoDB Ventures, Shield Capital, Harrison Chase, Bob van Luijt, Josh Lefkowitz
; source: Business Wire
; https://www.businesswire.com/news/home/20230719773647/en/Unstructured-Secures-%2425-Million-in-Seed-and-Series-A-Funding-to-Enable-Enterprises-to-Use-LLMs-With-their-Data
Company timeline
-
2026-07-01:
Active careers page with open commercial and public-sector roles
(hr)
. Unstructured’s careers page shows multiple open roles, including Senior Engineering Manager, Principal + Staff Software Engineers, Marketing Operations Manager, Staff Software Engineer, Site Reliability Engineer, Software Engineer - Public Sector, AI Engineer - Public Sector, and Navy Account Executive - Public Sector. ([unstructured.io](https://unstructured.io/careers?modal=try-for-free)). Team size observed: 51-200 employees employees.
Source: Unstructured careers.
https://unstructured.io/careers?modal=try-for-free
-
2026-06-09:
Repositioned around lakehouse-era unstructured data
(gtm)
. Unstructured published a broader positioning piece arguing that agents need a unified pipeline for unstructured data, not just structured lakehouse systems.
Source: Unstructured.
https://unstructured.io/blog/your-lakehouse-handles-structured-data-brilliantly-unstructured-is-next
-
2026-06-03:
Azure integration expands enterprise workflow connectivity
(product)
. Unstructured announced an expanded integration with Microsoft Azure to support enterprise AI workflows on Azure.
Source: Business Wire / Unstructured press.
https://www.businesswire.com/news/home/20260603131553/en/Unstructured-Expands-Integration-with-Microsoft-Azure-to-Power-Enterprise-AI-Workflows
-
2026-04-15:
Launched webhooks for downstream automation
(gtm)
. Unstructured added webhooks so completed jobs can trigger external systems and downstream processes automatically.
Source: Unstructured.
https://unstructured.io/blog/webhooks-connect-unstructured-to-everything-that-comes-after
-
2026-04-15:
Webhooks connect Unstructured jobs to downstream systems
(product)
. Unstructured added webhooks that fire on job lifecycle events such as scheduled, started, completed, stopped, and failed.
Source: Unstructured blog.
https://unstructured.io/blog/webhooks-connect-unstructured-to-everything-that-comes-after
-
2026-03-26:
Added Extract for database-oriented document workflows
(gtm)
. Unstructured introduced Extract as a new enrichment node for workflows that need records instead of chunks, broadening use cases beyond retrieval.
Source: Unstructured.
https://unstructured.io/blog/introducing-extract
LlamaIndex
Website: llamaindex.ai
LlamaIndex is an AI data framework and platform that helps developers and enterprises ingest, parse, transform, index, and query proprietary data for LLM and agent applications.
Category: Data Ingestion & Transformation.
Funding stage: Series A.
Founded: 2023.
Country: United States.
Employees: 11-50 employees.
Landscape fit: It fits this category because its core products center on converting unstructured enterprise documents and live data sources into AI-ready context through ingestion, parsing, transformation, and retrieval pipelines.
Funding rounds
-
Series A
announced 2025-03-01
: 19000000 USD
; investors: Norwest Venture Partners, Greylock
; source: PR Newswire
; https://www.prnewswire.com/news-releases/llamaindex-secures-19-million-series-a-to-power-enterprise-grade-knowledge-agents-302390936.html
-
Seed
announced 2023-05-01
: 8500000 USD
; investors: Greylock, Jack Altman, Lenny Rachitsky, Mathilde Collin, Raquel Urtasun, Joey Gonzalez
; source: Business Wire
; https://www.businesswire.com/news/home/20230531005251/en/LlamaIndex-Raises-%248.5M-to-Unlock-Large-Language-Models-Capabilities-with-Personal-Data
Company timeline
-
2026-06-01:
Hiring cluster announced across product, sales, engineering, and GTM
(hr)
. LlamaIndex announced multiple new team members in one LinkedIn post, including product marketing, solutions architecture, growth marketing, account executive, infrastructure engineering, software engineering, and applied research internships.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/llamaindex_just-a-tad-late-but-absolutely-worth-celebrating-activity-7461131945276542976-TfQQ
-
2026-05-01:
New teammates welcomed: Arthur Lorente and Simon Villanueva
(hr)
. LlamaIndex publicly welcomed two new teammates in a company LinkedIn post.. Team size observed: 11-50 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/llamaindex_were-thrilled-to-introduce-our-newest-teammates-activity-7439440520231383040-kHnQ
-
2026-02-26:
Customer-proofed document ingestion at scale with StackAI
(gtm)
. Hosted a webinar with StackAI on using LlamaCloud for high-accuracy document ingestion and retrieval across millions of enterprise files.
Source: LlamaIndex webinar landing page.
https://landing.llamaindex.ai/webinar-stackai-and-llamaindex-scaling-document-ingestion
-
2025-08-12:
Verticalized GTM for finance document agents
(gtm)
. Ran a finance-focused webinar promoting LlamaCloud parsing, extraction, and agentic orchestration for SEC filings and compliance workflows.
Source: LlamaIndex webinar landing page.
https://landing.llamaindex.ai/2025-aug-webinar
-
2025-03-01:
Series A
(funding)
. Amount: 19000000 USD. Investors: Norwest Venture Partners, Greylock
Source: PR Newswire.
https://www.prnewswire.com/news-releases/llamaindex-secures-19-million-series-a-to-power-enterprise-grade-knowledge-agents-302390936.html
-
2024-08-01:
Workflows beta replaced graph-style orchestration focus
(product)
. LlamaIndex introduced workflows beta as a new orchestration abstraction and explicitly framed it as a step beyond Query Pipelines for complex AI apps.
Source: Introducing workflows beta.
https://www.llamaindex.ai/blog/introducing-workflows-beta-a-new-way-to-create-complex-ai-applications-with-llamaindex
Reducto
Website: reducto.ai
Reducto is an AI document intelligence platform that parses complex documents and unstructured files into structured, AI-ready data for downstream workflows and LLM applications.
Category: Data Ingestion & Transformation.
Funding stage: Series B.
Founded: 2023.
Country: United States.
Employees: 78 employees.
Landscape fit: It fits Data Ingestion & Transformation because its core product converts proprietary documents into structured, reliable outputs that can be synchronized into RAG, automation, and other AI systems.
Funding rounds
-
Series B
announced 2025-10-01
: 75000000 USD
; investors: Andreessen Horowitz, Benchmark, First Round Capital, Y Combinator, BoxGroup
; source: Reducto
; https://reducto.ai/blog/reducto-series-b-funding
-
Series A
announced 2025-04-01
: 24500000 USD
; investors: Benchmark, First Round Capital, BoxGroup, Y Combinator
; source: Reducto
; https://reducto.ai/blog/reducto-series-a-funding
-
Seed
announced 2024-10-01
: 8400000 USD
; investors: First Round Capital, Y Combinator, BoxGroup, SV Angel, Liquid 2 Ventures
; source: Reducto
; https://reducto.ai/blog/seed-round
-
Seed
announced 2024-01-01
: 500000 USD
; investors: Y Combinator
; source: Y Combinator
; https://www.ycombinator.com/companies/reducto
Company timeline
-
2026-05-21:
Classify endpoint adds upstream routing to the platform
(gtm)
. Reducto launched a Classify endpoint that identifies document type before processing, giving users cheaper routing and more control over document handling.
Source: Reducto blog.
https://reducto.ai/blog/reducto-classify-endpoint-api
-
2026-05-21:
Classify and Split were upgraded for workflow routing
(product)
. Reducto added a Classify endpoint for pre-routing documents and launched Deep Split for longer documents and high-category workflows.
Source: Reducto blog.
https://reducto.ai/blog
-
2026-05-01:
Welcomed Apurv to the ML/AI team
(hr)
. Reducto announced Apurv joining the ML/AI team, highlighting prior experience at Two Sigma and Stripe.. Team size observed: 78 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/reducto-ai_welcoming-apurv-to-our-mlai-team-he-joins-activity-7456009635217453056-dl9T
-
2026-05-01:
Welcomed Henry as Chief of Staff to the CEO
(hr)
. Reducto announced Henry as Chief of Staff to the CEO, citing prior bizops work at Kikoff and consulting/research experience.. Team size observed: 78 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/reducto-ai_were-excited-to-welcome-henry-to-the-team-activity-7444497664597188608-56hO
-
2026-05-01:
Reducto acquired by Reducto
(acquisition)
. Reducto announced it is acquiring Opennote.. Acquirer: Reducto. Status: announced
Source: Reducto.
https://reducto.ai/blog/reducto-acquires-opennote
-
2026-04-06:
Deep Extract and Smart Schema introduced
(product)
. Reducto launched Deep Extract for stronger structured extraction and added Smart Schema in Studio to help teams create and improve schemas automatically.
Source: Reducto blog.
https://reducto.ai/blog
Ragie
Website: ragie.ai
Ragie is a managed context/retrieval infrastructure company that helps developers build AI applications on top of their own data with document parsing, ingestion, indexing, retrieval, and syncing across enterprise sources.
Category: Data Ingestion & Transformation.
Funding stage: Seed.
Founded: 2024.
Country: United States.
Employees: 2-10 employees.
Landscape fit: It fits Data Ingestion & Transformation because its core product parses, extracts, indexes, and continuously syncs proprietary documents and enterprise data into AI-ready context before retrieval or memory layers.
Funding rounds
-
Seed
announced 2024-08-01
: 5500000 USD
; investors: Craft Ventures, Valor, Saga VC, Chapter One
; source: Ragie
; https://www.ragie.ai/blog/intoducing-ragie-fully-managed-rag-as-a-service
Company timeline
-
2026-06-05:
Connector sync filters add finer ingestion control
(product)
. Ragie added glob-based sync filters so connectors can exclude documents by metadata pattern.
Source: Ragie changelog.
https://www.ragie.ai/changelog
-
2026-05-05:
MCP Bridge becomes generally available
(product)
. Ragie moved MCP Bridge from early access to general availability.
Source: Ragie changelog.
https://www.ragie.ai/changelog
-
2026-03-06:
Documents Elements API exposes structured extraction output
(product)
. Ragie added a Documents Elements API that returns structured elements such as titles, tables, images, and code blocks in reading order.
Source: Ragie changelog.
https://www.ragie.ai/changelog
-
2026-02-12:
Ragie Parse and web crawler API launch
(product)
. Ragie introduced Ragie Parse early access with Agentic OCR and made the web crawler connector manageable through the public API.
Source: Ragie changelog.
https://www.ragie.ai/changelog
-
2025-12-01:
Lead Infrastructure Engineer opening
(hr)
. Bob Remeika posted that Ragie was hiring a Lead Infrastructure Engineer in San Francisco and described the role as critical for scaling the platform.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/jackvirag_now-hiring-lead-infrastructure-engineer-activity-7396413842819473408-jKd7
-
2025-10-03:
Enterprise trust and compliance became a formal GTM pillar
(gtm)
. Ragie announced SOC 2 Type II, GDPR, and HIPAA compliance, along with partition isolation and demo-led enterprise messaging, making trust a central enterprise buying argument. ([ragie.ai](https://www.ragie.ai/blog/agent-ready-launch-week---day-5-enterprise-grade-security-and-compliance?utm_source=openai))
Source: Ragie blog / Base Chat page.
https://www.ragie.ai/blog/agent-ready-launch-week---day-5-enterprise-grade-security-and-compliance
Contextual AI
Website: contextual.ai
Contextual AI builds an enterprise context layer that ingests, syncs, and transforms private documents and data into AI-ready context for building grounded agents and applications.
Category: Data Ingestion & Transformation.
Founded: 2023.
Country: United States.
Employees: 51-200 employees.
Landscape fit: It fits Data Ingestion & Transformation because the product focuses on connecting to enterprise data sources, incrementally syncing content, extracting relevant context, and preparing it for retrieval and agent workflows before it reaches downstream AI systems.
Funding rounds
-
Series A
announced 2024-08-01
: 80000000 USD
; investors: Greycroft, Bain Capital Ventures, Lightspeed, Lip-Bu Tan, Conviction/Sarah Guo, Recall Capital, Bezos Expeditions, NVentures, HSBC Ventures, Snowflake Ventures
; source: Contextual AI
; https://contextual.ai/blog/announcing-series-a
-
Seed
announced 2023-06-01
: 20000000 USD
; investors: Bain Capital Ventures, Lightspeed, Greycroft, SV Angel, Elad Gil, Lip-Bu Tan, Sarah Guo, Amjad Masad, Harry Stebbings, Fraser Kelton, Sarah Niyogi, Nathan Benaich
; source: Contextual AI
; https://contextual.ai/blog/announcing-our-20m-seed-round-to-build-the-next-generation-of-language-models
Company timeline
-
2026-07-13:
LinkedIn jobs page shows no open roles
(hr)
. Contextual AI’s LinkedIn jobs page currently states that there are no jobs posted at the company.. Team size observed: 51-200 employees employees.
Source: LinkedIn.
https://www.linkedin.com/company/contextualai/jobs
-
2026-07-10:
Careers page remains active but with no visible postings
(hr)
. Contextual AI’s careers page shows an Open roles section and recruiting-oriented messaging, but no specific roles are visible in the page snapshot.. Team size observed: 51-200 employees employees.
Source: Contextual AI Careers.
https://contextual.ai/careers
-
2026-05-19:
Co-founder/CEO Douwe Kiela joins Google DeepMind
(hr)
. Bloomberg reported that Google DeepMind reached a deal to hire more than 20 researchers from Contextual AI, including co-founder and CEO Douwe Kiela.. Team size observed: 51-200 employees employees.
Source: Bloomberg Law.
https://news.bloomberglaw.com/legal-ops-and-tech/google-hires-staff-from-ai-startup-contextual-in-licensing-deal
-
2026-02-10:
Agent Composer public preview and demo datastores
(product)
. Contextual AI introduced Agent Composer in public preview, then added global datastores and web-search attributions for easier testing and app integration.
Source: Contextual AI documentation.
https://docs.contextual.ai/release-notes/2026
-
2026-01-26:
Agent Composer launch moves into orchestration and workflow automation
(gtm)
. Contextual AI launched Agent Composer, adding multi-tool orchestration, multi-step reasoning, static workflows, and enterprise-scale runtime for more complex technical work.
Source: Contextual AI blog.
https://contextual.ai/blog/introducing-agent-composer
-
2025-11-06:
Production evaluation workflow added
(product)
. Contextual AI built user feedback and annotation directly into the platform for collecting, categorizing, visualizing, and exporting human evaluation data.
Source: Contextual AI blog.
https://contextual.ai/blog/user-feedback-and-annotation
Holofin
Website: holofin.ai
Holofin is a Paris-based document intelligence platform that turns financial and administrative PDFs into validated, structured data for enterprise workflows.
Category: Data Ingestion & Transformation.
Founded: 2025.
Country: France.
Employees: 2-10 employees.
Landscape fit: It fits the AI context and data infrastructure category because it parses, extracts, validates, and synchronizes proprietary documents into AI-ready structured data before downstream systems use it.
Company timeline
-
2026-06-27:
Published extraction benchmark and performance proof
(gtm)
. Holofin launched a public bank-statement extraction benchmark claiming 98% of statements cleared with zero errors, using benchmark content to validate product reliability.
Source: Holofin blog.
https://holofin.ai/blog/
-
2026-06-01:
Enterprise packaging and security posture clarified
(product)
. Holofin’s pricing and trust pages now show pay-as-you-go and custom plans, API access, custom webhooks, EU data residency, RBAC, SSO/SAML, and on-premise deployment as part of the commercial stack.
Source: Pricing / Trust Center.
https://holofin.ai/pricing/
-
2026-06-01:
Document workflows and enterprise controls became explicit product surfaces
(product)
. Holofin’s product pages now present a chained workflow of classification, segmentation, extraction, validation, human review, export, and monitoring.
Source: Document Processing Workflows.
https://holofin.ai/capabilities/workflows/
-
2026-03-23:
Introduced document fraud detection as a product capability
(gtm)
. Holofin published fraud-detection content and a dedicated capability page, moving into forensic verification for tampered PDFs and financial documents.
Source: Holofin blog and capabilities page.
https://holofin.ai/blog/document-fraud-detection-what-a-pdf-cant-hide/
-
2026-03-23:
Fraud detection added as a native workflow step
(product)
. Holofin launched forensic document fraud detection with structural, typography, metadata, media, and security checks integrated into the pipeline.
Source: Document Fraud Detection: What a PDF Can't Hide.
https://holofin.ai/blog/document-fraud-detection-what-a-pdf-cant-hide/
-
2026-03-01:
Holofin launch announced by founder
(hr)
. Gregory Tappero introduced Holofin and said the company was live with customers running production workloads.. Team size observed: 2-10 employees employees.
Source: LinkedIn.
https://www.linkedin.com/posts/gtappero_introducing-holofinai-very-few-people-activity-7424388845128806400-AJGk
