Robotics Training Data

29 Companies 3 Categories Physical AI

Companies that capture human and robot demonstrations to produce the real-world data needed to train physical AI systems.

Filter and ordering

Order companies by Show category

Human Centric Demonstration Data

16 companies

Companies that record people performing physical tasks so robotics models can learn from natural human behavior in real-world environments.

Nferent AI

nferent.ai

Nferent AI builds Physical AI data infrastructure by capturing, labeling, and delivering real-world human skill datasets for robotics and humanoid AI companies, using multi-modal collection across factories, workshops, warehouses, and similar environments. ([nferent.ai](https://nferent.ai/?utm_source=openai))

Unknown Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Weak

Confidence: Low

EgoScale

egoscale.com

EgoScale is a robotics data company that provides large-scale, real-world egocentric human demonstrations and structured training data for VLA models, world models, and embodied AI.

Unknown Employees Seed Stage 2024 Founded

Trajectory Signals Strength Slight

Confidence: Low

Roborecs

roborecs.com

Roborecs is a Sofia-based company building EU-consented first-person human demonstration datasets for humanoid robot training, using head-mounted and wrist cameras plus multimodal sensors to capture physical work from the operator’s perspective.

Unknown Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Slight

Confidence: Low

Renlei Labs

renleilabs.com

Renlei Labs builds and sells egocentric human activity datasets for robotics and AI, capturing first-person video, motion, and sensor data from real workers performing real tasks.

Unknown Employees Unknown Stage Unknown Founded

Trajectory Signals Strength Slight

Confidence: Medium

Unidata

unidata.pro

Unidata is an AI training data company that provides data collection, expert labeling, and ready-made datasets, including egocentric first-person video and sensor data for robotics, AR/VR, and foundation models.

51-200 employees Employees Unknown Stage 2016 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Luel

luel.ai

Luel is a rights-cleared multimodal training-data marketplace and collection engine that supplies custom and off-the-shelf datasets to frontier AI teams.

11-50 employees Employees Seed Stage 2025 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Claru (Reka AI, Inc.)

claru.ai

Claru is a training-data platform for physical AI that collects and enriches egocentric video, robotics trajectories, world-model footage, and expert human judgments for robotics and embodied-AI teams.

11-50 employees Employees Series B Stage 2023 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Build AI

buildartificialintelligence.com

Build AI says it is building "high quality egocentric datasets for robotics & physical AI," including what it calls the largest egocentric data collection effort in history. ([buildartificialintelligence.com](https://www.buildartificialintelligence.com/))

2-10 employees Employees Unknown Stage 2025 Founded

Trajectory Signals Strength Slight

Confidence: Low

FPV Labs

fpvlabs.ai

FPV Labs is a research and product company building infrastructure for embodied AI, including Stera, a platform for capturing, processing, evaluating, and exporting egocentric multimodal human demonstration data for robot learning.

Unknown Employees Pre Seed Stage 2025 Founded

Trajectory Signals Strength Slight

Confidence: Medium

Neocambrian AI

neocambrian.ai

Neocambrian AI builds data-collection infrastructure for physical AI/robotics, capturing large-scale human action data to train embodied AI systems.

2-10 employees Employees Pre Seed Stage 2025 Founded

Trajectory Signals Strength Slight

Confidence: Medium

Human Archive

humanarchive.ai

Human Archive is a data lab that collects, labels, and synchronizes multimodal human behavior data to train robotics and physical-AI models. ([humanarchive.ai](https://www.humanarchive.ai/?utm_source=openai))

2-10 employees Employees Series A Stage 2026 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Hub

hub.xyz

Hub is a distributed real-world data infrastructure company that sources and delivers rights-cleared multimodal training data for frontier AI labs and robotics companies.

11-50 employees Employees Seed Stage 2024 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Cortex AI

cortexrobot.ai

Cortex AI builds large-scale real-world robot and egocentric human datasets for embodied AI, including workplace first-person video, robot trajectories, and human-in-the-loop evaluations.

3 Employees Seed Stage 2025 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Instawork

instawork.com

Instawork is a staffing marketplace that connects businesses with hourly workers, and it now also markets a robotics initiative that pays workers to record real-world task demonstrations for robot training.

501-1,000 employees Employees Series D+ Stage 2015 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Praxis AI

praxisrobotics.io

Praxis captures egocentric video, 3D scans, and other multimodal data from real industrial and residential environments to supply training data for physical AI and humanoid robotics.

Unknown Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Slight

Confidence: Medium

6thSense

6thsense.dev

6thSense builds tactile capture hardware and collects multimodal human demonstration data—combining touch, vision, depth, force, and hand pose—to help train dexterous robots.

2-10 employees Employees Pre Seed Stage 2026 Founded

Trajectory Signals Strength Slight

Confidence: Medium

Robot Centric Demonstration & Teleoperation Data

11 companies

Companies that capture tasks performed by robots under human guidance so models can learn directly from robot-native actions and trajectories.

DeepHow

deephow.com

DeepHow is an AI-powered operational knowledge platform that captures expert work, turns it into step-by-step digital instructions, and verifies execution for manufacturing, field service, and other frontline operations.

51-200 employees Employees Series A Stage 2018 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Scale AI

scale.com

Scale AI provides high-quality training data, evaluations, and applied AI tools, including a Physical AI Data Engine for robotics and other real-world AI systems.

501-1,000 employees Employees Unknown Stage 2016 Founded

Trajectory Signals Strength Strong

Confidence: Medium

OpenGraph Labs

opengraphlabs.com

OpenGraph Labs builds hardware and software for high-fidelity robotics data, including multimodal capture systems and tactile gloves for contact-rich robot learning.

2-10 employees Employees Unknown Stage 2025 Founded

Trajectory Signals Strength Slight

Confidence: Medium

Lightwheel

lightwheel.ai

Lightwheel is a physical AI infrastructure company that builds SimReady assets, egocentric human data, and evaluation platforms to help robotics teams train, validate, and deploy robot policies at scale.

201-500 employees Employees Series A Stage 2023 Founded

Trajectory Signals Strength Strong

Confidence: High

RLWRLD Inc.

rlwrld.ai

RLWRLD builds manipulation foundation models and a robotics intelligence layer that train industrial robots to handle dexterous, human-like tasks using real-world operational data.

51-200 employees Employees Unknown Stage 2023 Founded

Trajectory Signals Strength Strong

Confidence: High

Shift

joinshift.com

Shift is a robotics training-data company that pays people to record everyday household and trade tasks, then turns those recordings into anonymized datasets for AI and robotics customers.

Unknown Employees Seed Stage 2025 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

XDOF, Inc.

xdof.ai

XDOF builds data pipelines, collection tools, annotation systems, and teleoperation-driven datasets to help robotics and frontier AI labs train robots for physical-world tasks.

11-50 employees Employees Unknown Stage 2024 Founded

Trajectory Signals Strength Strong

Confidence: High

Mecka AI

mecka.ai

Mecka AI builds the data and deployment layer for physical AI, collecting and structuring real-world human activity data so robotics and embodied-AI teams can train and deploy systems more reliably.

11-50 employees Employees Series A Stage 2024 Founded

Trajectory Signals Strength Strong

Confidence: High

Sensei Robotics

senseirobotics.com

Sensei Robotics builds robotic training data infrastructure, connecting robotics companies to human operators who collect in-the-wild demonstration data for robot learning.

2-10 employees Employees Pre Seed Stage 2024 Founded

Trajectory Signals Strength Weak

Confidence: Low

Humanola

humanola.com

Humanola is a robotics data infrastructure company that provides a teleoperation platform and SDK to capture, sync, and convert robot demonstrations into training-ready datasets for embodied AI.

2-10 employees Employees Seed Stage 2025 Founded

Trajectory Signals Strength Moderate

Confidence: Medium

Aktoria Robotics

aktoria.com

Aktoria Robotics builds remote robot teleoperation and data-collection software that lets humans operate robots anywhere and capture robot-native demonstrations for training and fleet support.

2 Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Weak

Confidence: Low

Training Data Curation & Verification

2 companies

Companies that evaluate robot demonstrations and rollouts so raw recordings can become reliable training data for robotics models.

Instance

instancelabs.ai

Instance builds automated evaluation software for robot policies, starting with a verifier that judges whether robot rollouts succeeded and breaks tasks into subtasks from video data.

2-10 employees Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Slight

Confidence: Low

Shotwell

shotwell.ai

Shotwell is a robotics observability and data-quality company that helps robotics teams detect failures in video from both training data and live deployments, producing fast annotations for model evaluation and verification. ([ycombinator.com](https://www.ycombinator.com/companies/shotwellai?utm_source=openai))

2-10 employees Employees Unknown Stage 2026 Founded

Trajectory Signals Strength Slight

Confidence: Medium

Robotics Training Data competitive landscape summary

Robotics Training Data contains 29 companies across 3 categories. Companies that capture human and robot demonstrations to produce the real-world data needed to train physical AI systems.

Human Centric Demonstration Data

Companies that record people performing physical tasks so robotics models can learn from natural human behavior in real-world environments.

Nferent AI

Website: nferent.ai

Category: Human Centric Demonstration Data. Founded: 2026. Country: India.

Landscape fit: Captures human demonstrations in real environments to create training data for robotics and embodied AI.

Company timeline

2026-07-19: Founder publicly launched Nferent AI (hr) . Aman Jain stated in a Startup Pedia feature that he launched Nferent AI to solve the physical-data gap for robotics and AI. Source: Startup Pedia / LinkedIn. https://www.linkedin.com/posts/amanjain-unboxindustry_entrepreneurship-automation-ai-activity-7452316983515176960-q3Cj
2026-07-15: LiDAR and camera fusion emphasized for spatial AI capture (product) . Nferent AI published a sensor-selection post framing LiDAR and cameras as complementary inputs and tying the company’s methodology to multimodal perception data for robotics teams. Source: Nferent AI. https://nferent.ai/sensor-for-robot/
2026-07-15: Multi-modal sensor fusion and international market framing broaden (gtm) . Nferent AI’s July sensor-focused content framed the company as a multi-modal data collection provider serving robotics and Physical AI teams in the United States, China, and South Korea, while emphasizing LiDAR-plus-camera fusion. Source: Nferent AI. https://nferent.ai/sensor-for-robot/
2026-07-14: Founder highlighted Nferent AI Lab buildout (hr) . Aman Jain posted that Nferent AI Lab is building the data engine for Physical AI and embodied intelligence. Source: Aman Jain on LinkedIn. https://www.linkedin.com/posts/amanjain-unboxindustry_robot-physicalai-dataengine-activity-7472984105526185984-cCZi
2026-07-04: Teleoperation added as a core capture modality (product) . Nferent AI published a teleoperation-focused post that positioned remote operation as a backbone for high-quality robotics data collection and imitation-learning pipelines. Source: Nferent AI. https://nferent.ai/sensor-for-robot/
2026-07-04: Teleoperation is positioned as a quality differentiator (gtm) . A July post elevated teleoperation from a technical tactic to a central part of Nferent AI’s data-collection methodology for high-quality robotics datasets. Source: Nferent AI. https://nferent.ai/category/physical-ai-data-collection/

EgoScale

Website: egoscale.com

EgoScale is a robotics data company that provides large-scale, real-world egocentric human demonstrations and structured training data for VLA models, world models, and embodied AI.

Category: Human Centric Demonstration Data. Funding stage: Seed. Founded: 2024. Country: United States.

Landscape fit: Provides real-world egocentric human demonstrations for robotics and embodied AI training.

Company timeline

2026-02-18: Public launch of EgoScale dexterity scaling framework (gtm) . EgoScale published its human-to-robot transfer work on large-scale egocentric human video, showing a two-stage pretraining and mid-training recipe for dexterous manipulation and presenting the company as a provider of robot-ready egocentric data. Source: arXiv. https://arxiv.org/abs/2602.16710

Roborecs

Website: roborecs.com

Category: Human Centric Demonstration Data. Founded: 2026. Country: Bulgaria.

Landscape fit: Captures first-person human demonstrations for humanoid and robotics training.

Company timeline

2026-07-13: Skill-primitives framing for pre-launch capture design (gtm) . By mid-July, Roborecs said it was designing its pre-launch capture facility around modular skill primitives, showing that the product and collection pipeline were being shaped for repeatable dataset licensing. ([roborecs.com](https://roborecs.com/blog/robot-skill-primitives?utm_source=openai)) Source: Roborecs blog. https://roborecs.com/blog/robot-skill-primitives
2026-07-13: Capture pipeline redesigned around skill primitives (product) . Roborecs published a product post saying its pre-launch facility is being designed around modular skill primitives, with segmented demonstrations and boundary annotations rather than continuous unstructured clips. Source: Roborecs blog. https://roborecs.com/blog/skill-primitives-building-complex-tasks-from-reusable-parts
2026-07-04: Factory-floor narrative tied deployments to data capture (gtm) . Roborecs published a piece arguing that humanoid factory pilots are really data programs, reinforcing the idea that real-world deployments should feed the next training corpus. ([roborecs.com](https://roborecs.com/blog/humanoid-robots-in-manufacturing-2026?utm_source=openai)) Source: Roborecs blog. https://roborecs.com/blog/humanoid-robots-in-manufacturing-2026
2026-07-03: Category-level proprietary-data thesis (gtm) . In early July, Roborecs published a flagship argument that humanoid hardware is converging and the durable moat has shifted to proprietary robot data. ([roborecs.com](https://roborecs.com/blog/every-humanoid-company-is-a-data-company?utm_source=openai)) Source: Roborecs blog. https://roborecs.com/blog/every-humanoid-company-is-a-data-company
2026-07-03: EU-jurisdiction packaging and dataset provenance made explicit (product) . Roborecs’ site formalized the product as consent-cleared, provenance-tracked training data with dataset cards, chain-of-custody, and explicit LeRobot v3 / HDF5 delivery formats. Source: Roborecs homepage. https://roborecs.com/
2026-06-24: Channel comparison content positioned Roborecs against teleoperation data vendors (gtm) . A June explainer contrasted teleoperation and human-video data, effectively defining Roborecs as a supplier of low-cost first-person demonstrations that can be retargeted into robot training. ([roborecs.com](https://roborecs.com/blog/human-video-vs-teleoperation-data?utm_source=openai)) Source: Roborecs blog. https://roborecs.com/blog/human-video-vs-teleoperation-data

Renlei Labs

Website: renleilabs.com

Renlei Labs builds and sells egocentric human activity datasets for robotics and AI, capturing first-person video, motion, and sensor data from real workers performing real tasks.

Category: Human Centric Demonstration Data. Country: unknown.

Landscape fit: Captures first-person video and motion data from people performing real-world tasks for robotics training.

Company timeline

2026-07-01: Mexico-specific partner page localizes commercial outreach (gtm) . Renlei publishes a Spanish-language Mexico page that targets manufacturers, warehouses, restaurants, construction firms, workshops, automotive, hotels, agriculture, retail, and technical labs for small approved pilot collaborations. Source: Renlei Labs Mexico page. https://renleilabs.com/mexico
2026-07-01: Contributor and host-workplace acquisition motion appears (gtm) . Renlei adds a dedicated join flow that asks people to contribute data, make introductions, lead regional operations, or host real-world projects, with paid pilots and WhatsApp contact as part of onboarding. Source: Renlei Labs join page. https://renleilabs.com/join
2026-07-01: Public website launches dataset/API positioning (gtm) . Renlei's homepage presents the company as an API for multimodal human activity data and emphasizes egocentric video, motion, sensor streams, and dense annotations from real workers across multiple industries. Source: Renlei Labs homepage. https://renleilabs.com/

Unidata

Website: unidata.pro

Category: Human Centric Demonstration Data. Founded: 2016. Country: United Arab Emirates. Employees: 51-200 employees.

Landscape fit: Collects first-person demonstrations in operational environments for robotics and embodied AI training.

Company timeline

2026-07-14: Dataset catalog expansion into broader search-driven verticals (gtm) . By June and July 2026, Unidata’s blog and dataset pages expanded into sourcing, curation, and category guides such as autonomous driving and retail datasets, suggesting a broader inbound funnel around ready-made datasets. Source: Unidata blog category pages. https://unidata.pro/blog/category/datasets/dataset-collections/
2026-07-01: Hindi transcription hiring push (hr) . Unidata said it recruited and validated 20 qualified Hindi transcribers within a week for a speech data labeling project.. Team size observed: 51-200 employees employees. Source: LinkedIn. https://www.linkedin.com/company/unidatapro/
2026-06-18: Broader robotics content engine for technical buyers (gtm) . In late May and mid-June 2026, Unidata published several in-depth robotics and robot-training guides that explain collection, annotation, and pipeline design for technical decision-makers. Source: Unidata blog. https://unidata.pro/blog/robot-training-data-guide/
2026-06-18: Unidata introduced egocentric robot-training production guidance (product) . A June 2026 robotics post described Unidata’s production pipeline for egocentric data collection, including warehouse and dark-kitchen capture, multi-camera rigs, stream synchronization, and simulation-oriented scanning. ([unidata.pro](https://unidata.pro/blog/egocentric-data-collection-for-robot-training/?utm_source=openai)) Source: Unidata Blog. https://unidata.pro/blog/egocentric-data-collection-for-robot-training/
2026-06-01: Homepage consolidated services into an end-to-end training-data platform (product) . By mid-2026, the homepage presented collection, annotation, robots training, LLM training, AI model check, and ready-made datasets as a single integrated offer. ([unidata.pro](https://unidata.pro/?utm_source=openai)) Source: Unidata homepage. https://unidata.pro/
2026-06-01: Robotics dataset catalog broadened around egocentric and manipulation data (product) . The robotics dataset page surfaced first-person video, simulation data, and structured robot-learning datasets as part of a larger robotics training-data catalog. ([unidata.pro](https://unidata.pro/robotics-training-data/?utm_source=openai)) Source: Unidata website. https://unidata.pro/robotics-training-data/

Luel

Website: luel.ai

Luel is a rights-cleared multimodal training-data marketplace and collection engine that supplies custom and off-the-shelf datasets to frontier AI teams.

Category: Human Centric Demonstration Data. Funding stage: Seed. Founded: 2025. Country: United States. Employees: 11-50 employees.

Landscape fit: Provides rights-cleared first-person human demonstration datasets for robotics and physical AI.

Funding rounds

Seed announced 2026-05-01 : 31200000 USD ; investors: Lightspeed Venture Partners, General Catalyst, SV Angel ; source: Lightspeed Venture Partners ; https://lsvp.com/stories/our-investment-in-luel-the-marketplace-for-multimodal-ai-training-data/
Seed announced 2026-01-01 : 125000 USD ; investors: Y Combinator ; source: Dealroom ; https://app.dealroom.co/companies/luel

Company timeline

2026-07-01: Repositioning toward a multimodal lab and continuous data engine (gtm) . A newer Luel site variant describes the company as a multimodal lab powered by a continuous data engine, broadening the brand narrative beyond a simple marketplace. Source: Luel.work. https://luel.work/
2026-05-01: Published enterprise egocentric accessibility dataset page (product) . Luel added an enterprise dataset page for first-person accessibility video with synchronized audio, OCR, transcripts, motion data, and downloadable manifests plus consent/QA documentation. Source: Luel enterprise dataset page. https://www.luel.ai/datasets/egocentric-accessibility-vision
2026-05-01: Enterprise and contributor packaging solidifies (gtm) . The main site adds clear dual-path packaging for enterprise dataset buyers and contributors, including catalog browsing, custom dataset requests, and earnings/payout flows. Source: Luel homepage. https://www.luel.ai/
2026-05-01: Seed (funding) . Amount: 31200000 USD. Investors: Lightspeed Venture Partners, General Catalyst, SV Angel Source: Lightspeed Venture Partners. https://lsvp.com/stories/our-investment-in-luel-the-marketplace-for-multimodal-ai-training-data/
2026-04-01: Contributor-network scale becomes a demand-side proof point (gtm) . Luel’s content and homepage messaging start to use a large contributor network and rapid fulfillment claims as proof that the marketplace can supply data at scale. Source: Luel blog and homepage. https://www.luel.ai/blog/10x-faster-data-collection-how-ai-training-data-providers-achieve-speed
2026-04-01: Dataset catalog expands into speech, video, and sensor use cases (gtm) . The enterprise catalog and category pages show Luel packaging off-the-shelf speech, video, and sensor datasets, including robotics- and accessibility-oriented collections. Source: Luel enterprise and dataset pages. https://www.luel.ai/enterprise

Claru (Reka AI, Inc.)

Website: claru.ai

Category: Human Centric Demonstration Data. Funding stage: Series B. Founded: 2023. Country: United States. Employees: 11-50 employees.

Landscape fit: Collects and enriches egocentric human demonstrations for physical AI and robotics models.

Funding rounds

Series B announced 2025-07-01 : 110000000 USD ; investors: NVIDIA, Snowflake ; source: Reka ; https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms
Series A announced 2023-06-01 : 60000000 USD ; investors: DST Global Partners, Radical Ventures, Snowflake Ventures, Nat Friedman ; source: Reka ; https://reka.ai/news/announcing-our-60m funding-to-build-generative-models-and-advance-ai-research
Series Seed announced 2022-09-01 : 7450000 USD ; source: Forge Global ; https://forgeglobal.com/reka-ai_ipo/

Company timeline

2026-06-01: Robotics-specific proof motion added through detailed case studies (gtm) . Claru surfaced case studies for egocentric robotics collection, workplace data capture, and game-based capture, each with scale metrics and operational details. Source: Claru. https://claru.ai/case-studies
2026-06-01: High-intent solution pages expanded around egocentric and physical-AI datasets (gtm) . Claru broadened its website into dedicated solution and dataset pages for egocentric video, embodied AI, teleoperation, robotics, and related physical-AI use cases. Source: Claru. https://claru.ai/egocentric-video-data
2026-05-26: Cluster of physical-AI capture jobs posted (hr) . Claru posted a large set of contractor roles on its jobs board, including household, retail, warehouse, office, hospitality, restaurant, indoor activity, garden, and garage capture work in the United States and Latin America.. Team size observed: 11-50 employees employees. Source: Claru AI Jobs. https://claru.ai/jobs
2026-05-04: Humanoid-robotics angle surfaced in editorial and solution messaging (gtm) . Claru began explicitly addressing humanoid robot training-data requirements, linking its collection network to the needs of humanoid labs and sim-to-real transfer. Source: Claru. https://claru.ai/blog
2026-04-16: Game data collection role posted (hr) . Claru posted a remote worldwide Game Data Collector opening for gameplay recording and human demonstration data collection.. Team size observed: 11-50 employees employees. Source: Claru AI Jobs. https://claru.ai/jobs/game-data-collector
2026-04-01: Technical thought-leadership program launched for robotics data buyers (gtm) . Claru published a sequence of educational posts on physical-AI training data, VLA requirements, sim-to-real gaps, and egocentric video providers to capture search demand from robotics engineers. Source: Claru. https://claru.ai/blog

Build AI

Website: buildartificialintelligence.com

Category: Human Centric Demonstration Data. Founded: 2025. Country: United States. Employees: 2-10 employees.

Landscape fit: Builds large-scale egocentric datasets from people performing physical tasks for robotics training.

Funding rounds

Seed announced 2025-09-01 : 5000000 USD ; investors: Abstract Ventures, Pear VC, HF0, ZFellows, Alex Botez ; source: StartupHub.ai ; https://www.startuphub.ai/ai-news/funding-round/2025/build-ais-5m-bet-on-human-view-robotics

Company timeline

2026-07-20: 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/buildai-compute
2026-07-01: Current website shifted to a broader visual-industrial presentation (product) . The company homepage now pairs the egocentric dataset positioning with an 'AI-Powered Manufacturing' experience that highlights interactive 3D visualizations of real factory environments. Source: Build AI website. https://www.buildartificialintelligence.com/
2026-06-04: Public customer-proof and credibility refresh on homepage (gtm) . The homepage began foregrounding named customers/investors and quantified claims such as time and cost savings and customer ROI. ([launchpadbuild.ai](https://www.launchpadbuild.ai/?utm_source=openai)) Source: Launchpad Build AI. https://www.launchpadbuild.ai/
2026-04-30: El Segundo HQ and rebrand to Launchpad Build AI (gtm) . The company announced a US headquarters in El Segundo and a rebrand to Launchpad Build AI alongside senior leadership additions. ([launchpadbuild.ai](https://www.launchpadbuild.ai/launchpad-build-ai-enters-new-growth-phase-with-el-segundo-hq-senior-hires-and-world-first-manufacturing-language-model/?utm_source=openai)) Source: Launchpad Build AI. https://www.launchpadbuild.ai/launchpad-build-ai-enters-new-growth-phase-with-el-segundo-hq-senior-hires-and-world-first-manufacturing-language-model/
2026-04-30: Productization around Manufacturing Language Model (gtm) . The company published a technology page centering its Manufacturing Language Model and Digitool robotics system, positioning itself as an AI company rather than a robotics integrator or consultancy. ([launchpadbuild.ai](https://www.launchpadbuild.ai/our-technology/?utm_source=openai)) Source: Launchpad Build AI. https://www.launchpadbuild.ai/our-technology/
2026-02-05: Strategic investor motion strengthened industrial credibility (gtm) . Launchpad Build AI announced PXN Ventures as a new investor, while highlighting a syndicate that included strategic backers such as Ericsson Ventures and Lockheed Martin Ventures. ([launchpadbuild.ai](https://www.launchpadbuild.ai/launchpad-announces-pxn-ventures-as-its-newest-strategic-investor/)) Source: Launchpad Build AI. https://www.launchpadbuild.ai/launchpad-announces-pxn-ventures-as-its-newest-strategic-investor/

FPV Labs

Website: fpvlabs.ai

Category: Human Centric Demonstration Data. Funding stage: Pre Seed. Founded: 2025. Country: India.

Landscape fit: Captures and processes multimodal first-person human demonstrations for robot learning.

Company timeline

2026-06-11: Published benchmark proof for commodity-hardware capture (gtm) . FPV Labs published a validation post comparing Stera's ARKit pose against a 24-camera Vicon system, using it to prove the fidelity of its iPhone-based capture stack. Source: How accurate is an iPhone tracking, really?. https://www.fpvlabs.ai/essays/how-accurate-is-an-iphone-really
2026-06-11: Validation of iPhone ARKit pose against Vicon (product) . FPV Labs published a benchmark comparing Stera’s ARKit pose estimates on an iPhone 12 Pro with a 24-camera Vicon system over hour-long sessions. Source: How accurate is an iPhone tracking, really?. https://www.fpvlabs.ai/essays/how-accurate-is-an-iphone-really
2026-05-14: Added self-serve access, docs, and open distribution channels (gtm) . FPV Labs exposed Stera through documentation, a platform/dashboard flow, a free tier, and Hugging Face dataset hosting, while reserving direct contact for larger organizations. Source: Stera documentation and product pages. https://www.fpvlabs.ai/stera
2026-05-14: Launched Project Stera as the core product and open dataset motion (gtm) . FPV Labs released Project Stera, combining an iOS capture app, processing SDK, evaluation layer, export tooling, and the open Stera-10M dataset on commodity hardware. Source: Introducing Project Stera. https://www.fpvlabs.ai/essays/launching-stera
2026-05-14: Stera documentation and processing stack go live (product) . The docs exposed the full workflow, including capture, dashboard review, Python processing, one-call evaluation reports, and episode export with multiple output formats. Source: Stera Documentation. https://www.fpvlabs.ai/stera/docs
2026-05-14: Project Stera launches with app, SDK, and Stera-10M (product) . FPV Labs released Project Stera as a full-stack embodied-AI data platform, including the Stera-10M dataset and the Stera Suite processing pipeline for training-ready exports. Source: Introducing Project Stera. https://www.fpvlabs.ai/essays/launching-stera

Neocambrian AI

Website: neocambrian.ai

Neocambrian AI builds data-collection infrastructure for physical AI/robotics, capturing large-scale human action data to train embodied AI systems.

Category: Human Centric Demonstration Data. Funding stage: Pre Seed. Founded: 2025. Country: India. Employees: 2-10 employees.

Landscape fit: Collects large-scale human action data to train embodied AI and robotics systems.

Company timeline

2026-07-01: Hiring two founders-office associates (hr) . Founder Abhinav Kukreja posted that Neocambrian AI was hiring two founders-office associates to work directly with him on the most important levers for the business.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/posts/abhinavkukreja_i-started-my-first-startup-at-17-now-i-am-activity-7477334006645071873-UW8v
2026-06-01: Team member joined from AMD Boston (hr) . Sparsh C. publicly announced moving back to India and joining Neocambrian AI after around 1.5 years at AMD Boston.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://in.linkedin.com/in/sparshchaudhri
2026-06-01: Public launch of Neocambrian AI as a physical-AI data infrastructure company (gtm) . The company publicly announced Neocambrian AI and framed it as building the data foundation for physical AI, including an India-based robotics data factory for large-scale human-action datasets. Source: LinkedIn / company web presence. https://www.linkedin.com/company/neocambrianai
2026-05-26: Launch of India-focused robotics data factory (product) . Neocambrian AI publicly launched a robotics data factory in India to collect large-scale human action datasets for Physical AI and embodied robotics systems. ([startupstorymedia.com](https://startupstorymedia.com/neocambrian-ai-launches-india-focused-robotics-data-factory-for-physical-ai-models/?utm_source=openai)) Source: StartupStory. https://startupstorymedia.com/neocambrian-ai-launches-india-focused-robotics-data-factory-for-physical-ai-models/
2026-04-01: Hiring Founding Researcher (Robotics) (hr) . A LinkedIn job post advertised a Founding Researcher (Robotics) role at Neocambrian AI in Noida, with a starting date of April 2026 and responsibility for robotics data/model infrastructure.. Team size observed: 2-10 employees employees. Source: LinkedIn Jobs. https://in.linkedin.com/jobs/view/founding-researcher-robotics-at-neocambrian-ai-4394256510
2026-04-01: Hiring for Head of Growth (hr) . Founder posted that Neocambrian AI was hiring someone to lead growth in Delhi NCR, with an immediate start and a paid referral incentive.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/posts/abhinavkukreja_were-hiring-someone-to-lead-growth-at-neocambrian-activity-7441447291917611008-51vi

Human Archive

Website: humanarchive.ai

Category: Human Centric Demonstration Data. Funding stage: Series A. Founded: 2026. Country: United States. Employees: 2-10 employees.

Landscape fit: It fits Human-Centric Demonstration Data because its core product is recorded human motion/task data—video, audio, tactile, mocap, and synchronized sensors—used to train robots and physical AI systems. ([humanarchive.ai](https://www.humanarchive.ai/?utm_source=openai))

Funding rounds

Seed announced 2026-05-01 : 8200000 USD ; investors: Wing Venture Capital, NVP Capital, Y Combinator, angel investors from OpenAI, angel investors from Nvidia, angel investors from Google, angel investors from Mercor, angel investors from AfterQuery, angel investors from BAIR, angel investors from SAIL, Brad Boa, angel investors from Meta ; source: Human Archive official website ; https://www.humanarchive.ai/

Company timeline

2026-07-01: Founder said the team was working seven days a week and hiring ML engineers (hr) . A company LinkedIn post said Human Archive’s young, lean research team in San Francisco was working seven days a week and invited interested ML engineers to reach out.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/company/human-archive
2026-07-01: Current careers page shows active hiring across 8 roles (hr) . The Human Archive website and YC company jobs page currently show an active hiring slate, including Head of Engineering, Industrial Designer, QA / Hardware Test Engineer, Firmware Engineer, Embedded / Electrical Engineer, Mechanical Engineer - Wearable Systems, Videographer, and Operations Manager.. Team size observed: 2-10 employees employees. Source: Human Archive / Y Combinator. https://www.humanarchive.ai/
2026-06-01: SF office opened and recruiting expanded (hr) . A founder post said Human Archive had just opened its San Francisco office and was looking for research engineers, hardware engineers, and ops generalists.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/company/human-archive
2026-05-26: Public launch of multimodal robotics dataset platform (gtm) . Human Archive launched its public GTM around "the world's largest multimodal robotics dataset," framing the company as infrastructure for embodied intelligence and explicitly marketing dataset products and customer samples to robotics and world-model teams. ([ycombinator.com](https://www.ycombinator.com/launches/PeP-human-archive-the-world-s-largest-multimodal-robotics-dataset)) Source: Y Combinator Launch YC. https://www.ycombinator.com/launches/PeP-human-archive-the-world-s-largest-multimodal-robotics-dataset
2026-05-26: Launch of HA-Multi and HA-Ego datasets (product) . Human Archive publicly launched as a multimodal robotics dataset company, introducing HA-Multi and HA-Ego plus a capture-and-processing stack built from custom hardware rigs, QA, anonymization, and annotation pipelines. ([ycombinator.com](https://www.ycombinator.com/launches/PeP-human-archive-the-world-s-largest-multimodal-robotics-dataset)) Source: Y Combinator Launch YC. https://www.ycombinator.com/launches/PeP-human-archive-the-world-s-largest-multimodal-robotics-dataset
2026-05-01: Seed (funding) . Amount: 8200000 USD. Investors: Wing Venture Capital, NVP Capital, Y Combinator, angel investors from OpenAI, angel investors from Nvidia, angel investors from Google, angel investors from Mercor, angel investors from AfterQuery, angel investors from BAIR, angel investors from SAIL, Brad Boa, angel investors from Meta Source: Human Archive official website. https://www.humanarchive.ai/

Hub

Website: hub.xyz

Hub is a distributed real-world data infrastructure company that sources and delivers rights-cleared multimodal training data for frontier AI labs and robotics companies.

Category: Human Centric Demonstration Data. Funding stage: Seed. Founded: 2024. Country: United States. Employees: 11-50 employees.

Landscape fit: Sources real-world multimodal human demonstrations for robotics and physical AI training.

Funding rounds

seed announced 2025-09-01 : 1700000 USD ; investors: SwissBorg ; source: Business Wire ; https://www.businesswire.com/news/home/20250925811310/en/Hub.xyz-Closes-Major-Funding-Round-with-SwissBorg-to-Scale-The-First-Distributed-AI-Data-Infrastructure-Network

Company timeline

2026-07-01: Brazil Operations & Logistics Lead job posted (hr) . A LinkedIn job post for Hub advertised a Brazil Operations & Logistics Lead role in the San Francisco Bay Area context, with the role centered on importing gear, customs, shipping, and contributor logistics in Brazil.. Team size observed: 11-50 employees employees. Source: LinkedIn Jobs. https://www.linkedin.com/jobs/view/brazil-operations-logistics-lead-at-hub-xyz-4429264599
2026-07-01: Brazil hiring push announced (hr) . Tim Sprecher posted that Hub was hiring in Brazil for a social media manager and a logistics operations manager, both based in São Paulo and hybrid.. Team size observed: 11-50 employees employees. Source: LinkedIn. https://www.linkedin.com/company/hub-xyz
2026-06-01: Brazil contributor-growth hiring mention (hr) . Tim Sprecher posted that Hub needed someone to build the Brazil operation, describing the Brazil Operator role and linking a Marketing Growth Operator opening in Greater São Paulo.. Team size observed: 11-50 employees employees. Source: LinkedIn. https://www.linkedin.com/company/hub-xyz
2026-06-01: Public launch on YC with frontier AI and robotics positioning (gtm) . Hub publicly launched on YC as a provider of real-world training data for frontier AI labs and robotics, citing a 7-figure revenue agreement and hiring robotics talent. ([ycombinator.com](https://www.ycombinator.com/launches/Qoe-hub-real-world-training-data-for-frontier-ai-labs-and-robotics)) Source: Y Combinator. https://www.ycombinator.com/launches/Qoe-hub-real-world-training-data-for-frontier-ai-labs-and-robotics
2026-05-01: Rolled out contributor waitlist and reward-based data training program (gtm) . Hub’s consumer-facing platform invited users to contribute images, audio, and video to train frontier models, with access opening in queue order and USDC rewards accruing to contributors. ([ai.hub.xyz](https://ai.hub.xyz/?utm_source=openai)) Source: Hub / LinkedIn. https://ai.hub.xyz/
2025-10-21: Opened U.S. headquarters in Palo Alto (gtm) . Hub announced a new U.S. headquarters in Palo Alto as part of its expansion into the American AI infrastructure market and Bay Area partner ecosystem. ([businesswire.com](https://www.businesswire.com/news/home/20251021183278/en/Hub.xyz-Expands-into-the-U.S.-with-New-Headquarters-in-Palo-Alto-California?utm_source=openai)) Source: Business Wire. https://www.businesswire.com/news/home/20251021183278/en/Hub.xyz-Expands-into-the-U.S.-with-New-Headquarters-in-Palo-Alto-California

Cortex AI

Website: cortexrobot.ai

Cortex AI builds large-scale real-world robot and egocentric human datasets for embodied AI, including workplace first-person video, robot trajectories, and human-in-the-loop evaluations.

Category: Human Centric Demonstration Data. Funding stage: Seed. Founded: 2025. Country: United States. Employees: 3.

Landscape fit: Captures egocentric workplace demonstrations for training robotics and physical AI systems.

Funding rounds

Seed VC announced 2025-11-01 : 6000000 USD ; investors: 500 Global, Gokul Rajaram, Golden Gate Ventures, Kaz Nejatian, Liquid 2 Ventures, Matrix Partners, Pioneer Fund, Spot VC, Y Combinator ; source: CB Insights ; https://www.cbinsights.com/company/cortexai/financials

Company timeline

2026-07-01: Open engineering opening on careers page (hr) . Cortex AI's careers page listed a Software Engineer / Intern opening for San Francisco and Singapore.. Team size observed: 3 employees. Source: Cortex AI careers page. https://cortexrobot.ai/careers/
2026-06-01: Expanded ops hiring into procurement and hardware infrastructure (gtm) . Cortex AI added a combined robot-operator/procurement role and a robotics hardware engineer role, indicating it is building the operational and mechanical infrastructure needed to repeatedly deliver bespoke datasets and evaluations. Source: Cortex AI Careers. https://cortexrobot.ai/careers/robot-operator-procurement
2026-06-01: Hiring for robot operators to scale data collection operations (gtm) . Cortex AI posted full-time robot-operator roles in Singapore focused on teleoperation, egocentric capture, annotation, and maintenance of the lab and sensors that support large-scale data collection. Source: Cortex AI Careers. https://cortexrobot.ai/careers/robot-operator
2026-06-01: Robot Operator hiring opening (hr) . Cortex AI posted a role for full-time Robot Operators to support teleoperation and egocentric data collection for embodied AI training.. Team size observed: 3 employees. Source: Cortex AI careers page. https://cortexrobot.ai/careers/robot-operator
2026-05-28: MolmoAct 2 release expanded the ecosystem around Cortex-supported data (product) . Ai2 released MolmoAct 2, open-sourcing the code, training data, and LeRobot integration after the initial model launch, which broadened downstream demand for the kind of datasets and evaluations Cortex provides. Source: Ai2 blog. https://allenai.org/blog/molmoact2
2026-05-05: Third-party validation work on MolmoAct 2 (gtm) . Ai2 said it retained Cortex AI to run independent real-world benchmarking of MolmoAct 2, positioning Cortex as a trusted evaluation and data partner for a high-profile robotics foundation model release. Source: Ai2 Blog. https://allenai.org/blog/molmoact2

Instawork

Website: instawork.com

Category: Human Centric Demonstration Data. Funding stage: Series D+. Founded: 2015. Country: United States. Employees: 501-1,000 employees.

Landscape fit: It fits this category because its robotics page and recent news indicate it is using its worker network to capture human task demonstrations and other real-world footage for training robotics models, which is human-centric demonstration data.

Funding rounds

Series D announced 2023-05-01 : 60000000 USD ; investors: TCV, 9Yards Capital, Larry Fitzgerald Jr., Benchmark, Spark Capital, Craft Ventures, Greylock, Four Rivers Group, GV ; source: Instawork ; https://www.instawork.com/blog/instawork-invests-in-artificial-intelligence-to-optimize-how-businesses-find-the-workers-they-need
Series C announced 2021-07-01 : 60000000 USD ; investors: Craft Ventures, Greylock, Corner Ventures, Four River Group, WndrCo, Tilman Fertitta, Benchmark, Spark Capital, GV, Burst Capital, SV Angel ; source: PR Newswire ; https://www.prnewswire.com/news-releases/instawork-raises-60m-to-rapidly-expand-its-work-marketplace-connecting-local-businesses-with-skilled-hourly-professionals-301327941.html
Series B announced 2019-05-01 : 18000000 USD ; investors: Spark Capital, GV, Burst Capital, Benchmark, Y Combinator, Tuesday Capital, SV Angel ; source: Instawork ; https://www.instawork.com/blog/new-markets-and-new-investors-in-pursuit-of-our-vision
Series A announced 2018-02-01 : 8200000 USD ; investors: Benchmark, Y Combinator, Tuesday Capital, Liquid 2 Ventures, SV Angel ; source: Global Venturing ; https://globalventuring.com/blog/2019/05/24/instawork-recruits-gv-for-18m-round/

Company timeline

2026-07-21: Current careers page shows a robotics and operations hiring cluster (hr) . Instawork’s careers page currently lists open roles including Robotics AI Campus Lead, Staff Software Engineer - Instawork Robotics, Senior Software Engineer - Instawork Robotics, Technical Partnerships Lead, General Manager, Data Centers, and multiple operations roles across the U.S., Canada, Mexico, and India.. Team size observed: 501-1,000 employees employees. Source: Instawork Careers. https://www.instawork.com/careers?gh_jid=4387998006
2026-04-16: Shifted into robotics as a new GTM surface (gtm) . Instawork launched a robotics page and accompanying blog that position the company as a provider of human data collection, certification, and field operations for embodied AI and robotics deployment. ([instawork.com](https://www.instawork.com/robotics)) Source: Instawork blog / Robotics landing page. https://www.instawork.com/robotics
2026-04-16: Instawork launches robotics lab for training robots and deploying field ops (product) . Instawork said it had turned its worker network into Instawork Robotics Lab, a platform for training robots, certifying workers for data collection and field ops, and connecting businesses that deploy automation. ([instawork.com](https://www.instawork.com/blog/human-advantage-in-the-robotics-revolution?utm_source=openai)) Source: Instawork blog + robotics page. https://www.instawork.com/blog/human-advantage-in-the-robotics-revolution
2025-04-24: Won retail workforce-hiring award (gtm) . Instawork was named Workforce Hiring Solution of the Year by RetailTech Breakthrough, reinforcing its retail positioning. ([instawork.com](https://www.instawork.com/blog/instawork-named-workforce-hiring-solution-of-the-year-by-retailtech-breakthrough?utm_source=openai)) Source: Instawork blog. https://www.instawork.com/blog/instawork-named-workforce-hiring-solution-of-the-year-by-retailtech-breakthrough
2025-04-03: Rolled out warehouse pay and staffing insights (gtm) . Instawork continued publishing warehouse-focused labor analysis, framing pay growth and trade-policy uncertainty as reasons operators should rethink staffing models. ([instawork.com](https://www.instawork.com/blog/warehouses-lead-march-pay-growth-as-businesses-brace-for-trade-policy-shifts?utm_source=openai)) Source: Instawork blog. https://www.instawork.com/blog/warehouses-lead-march-pay-growth-as-businesses-brace-for-trade-policy-shifts
2025-03-31: Instawork broadens into longer-term staffing and permanent hiring (product) . Instawork introduced long-term staffing, then launched Instawork Hiring for permanent roles with AI matching, shortlisted candidates, trial shifts, and enterprise-oriented hiring tiers. ([instawork.com](https://www.instawork.com/blog/instawork-introduces-long-term-staffing)) Source: Instawork blog + product page + help center. https://www.instawork.com/blog/instawork-introduces-long-term-staffing

Praxis AI

Website: praxisrobotics.io

Praxis captures egocentric video, 3D scans, and other multimodal data from real industrial and residential environments to supply training data for physical AI and humanoid robotics.

Category: Human Centric Demonstration Data. Founded: 2026. Country: United States.

Landscape fit: It fits this category because its core product is human- and environment-captured demonstration data used to train robotics models, rather than robots themselves.

Funding rounds

Seed announced 2026-06-01 : 125000 USD ; investors: Y Combinator ; source: Dealroom ; https://app.dealroom.co/companies/praxis_ai_1

Company timeline

2026-07-01: Out-of-stealth framing around physical-world data supply (gtm) . A founder post described Praxis as coming out of stealth to capture the physical-world data humanoid robots and frontier models learn from, reinforcing the company's data-supply thesis and broader environment coverage. Source: LinkedIn post by founder Rohan. https://www.linkedin.com/posts/rohan-s-1186b2214_were-bringing-praxis-yc-s26-out-of-stealth-activity-7478237308106752000-dnLP
2026-07-01: Shift to embodied-AI data capture and request-access GTM (gtm) . Praxis's current public site reframes the company around capturing egocentric video, 3D scans, and multimodal data from industrial and residential environments, with a request-access funnel and YC backing prominently displayed. Source: Praxis website. https://www.praxisrobotics.io/
2026-07-01: Public launch of multimodal physical-AI data capture product (product) . Praxis publicly surfaced as a YC Summer 2026 company with a product focused on capturing egocentric video, 3D scans, and other multimodal data from real industrial and residential environments for embodied AI and humanoid robotics. The homepage emphasizes an access request flow and live reconstruction / capture interactions, indicating an early controlled rollout. ([ycombinator.com](https://www.ycombinator.com/companies/praxis-ai-2)) Source: Y Combinator and Praxis homepage. https://www.ycombinator.com/companies/praxis-ai-2
2026-06-01: Seed (funding) . Amount: 125000 USD. Investors: Y Combinator Source: Dealroom. https://app.dealroom.co/companies/praxis_ai_1
2026-02-01: YC launch as manufacturing reasoning-agent platform (gtm) . Praxis launched on Y Combinator as a manufacturing-specific AI copilot for predicting and preventing machine failures, with messaging around root-cause analysis, shop-floor workflows, and no-code operational ML tools. Source: Y Combinator Launch YC. https://www.ycombinator.com/launches/MKz-praxis-ai-reasoning-agents-fine-tuned-for-manufacturing

6thSense

Website: 6thsense.dev

6thSense builds tactile capture hardware and collects multimodal human demonstration data—combining touch, vision, depth, force, and hand pose—to help train dexterous robots.

Category: Human Centric Demonstration Data. Funding stage: Pre Seed. Founded: 2026. Country: United States. Employees: 2-10 employees.

Landscape fit: It fits this category because it records people performing real physical tasks with wearable sensing rigs in real environments, producing human-centric demonstration data for robotics training.

Funding rounds

Pre-Seed announced 2026-07-01 : 1000000000 KRW ; investors: Krew Capital ; source: VentureSquare ; https://www.venturesquare.net/1097147/

Company timeline

2026-07-01: Expanded modality and labeling scope on homepage (product) . The homepage broadened the declared capture schema to eight aligned modalities, including RGB, depth, hand pose, IMU dynamics, secondary cameras, dense commentary, and outcome labels. Source: 6thSense homepage. https://6thsense.dev/
2026-07-01: YC launch and founder lineup surfaced (hr) . 6thSense appeared on Y Combinator as an active Summer 2026 company with founders James Baek, Ronak Agarwal, Alex Hyungwoo Noh, and Matt Wulff.. Team size observed: 2-10 employees employees. Source: Y Combinator. https://www.ycombinator.com/companies/6thsense
2026-07-01: Public launch of tactile capture platform for dexterous robots (gtm) . 6thSense publicly launched as a vendor of wearable tactile sensing rigs and multimodal demonstration data for dexterous robot learning, framing the product as a ready-made capture stack for touch, vision, depth, force, and hand pose. Source: Y Combinator. https://www.ycombinator.com/launches/R3X-6thsense-giving-robots-touch
2026-07-01: Public launch of tactile capture stack positioning (product) . 6thSense publicly presented its product as tactile capture hardware for dexterous robotics, centered on a glove, tactile skin, and egocentric rig that capture aligned multimodal human demonstrations. Source: 6thSense product page. https://6thsense.dev/product
2026-07-01: Pre-Seed (funding) . Amount: 1000000000 KRW. Investors: Krew Capital Source: VentureSquare. https://www.venturesquare.net/1097147/

Robot Centric Demonstration & Teleoperation Data

Companies that capture tasks performed by robots under human guidance so models can learn directly from robot-native actions and trajectories.

DeepHow

Website: deephow.com

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Series A. Founded: 2018. Country: United States. Employees: 51-200 employees.

Landscape fit: Captures skilled physical work and turns demonstrations into structured data for operational learning systems.

Funding rounds

Convertible Note announced 2025-05-01 : 5000000 USD ; source: CB Insights ; https://www.cbinsights.com/company/deephow/financials
Series A announced 2023-04-01 : 14000000 USD ; investors: Owl Ventures, LG Technology Ventures, Sierra Ventures, Qualcomm Ventures, Osage Venture Partners, Foothill Ventures ; source: PR Newswire ; https://www.prnewswire.com/news-releases/deephow-raises-14m-series-a-funding-to-transform-how-enterprises-capture-and-transfer-technical-know-how-301809091.html
Pre-Series A announced 2021-09-01 : 9000000 USD ; investors: Sierra Ventures, Osage Venture Partners, Qualcomm Ventures LLC, Foothill Ventures ; source: PR Newswire ; https://www.prnewswire.com/news-releases/deephow-raises-9m-pre-series-a-funding-to-bridge-the-skills-gap-in-the-manufacturing-service-and-construction-industries-301377006.html
Seed announced 2018-07-01 : 120000 USD ; investors: Fontinalis Partners, Right Side Capital Management ; source: Dealroom.co ; https://app.dealroom.co/companies/deephow

Company timeline

2026-07-19: Active AI Engineer opening posted (hr) . DeepHow has an active LinkedIn job posting for an Artificial Intelligence Engineer in the United States.. Team size observed: 51-200 employees employees. Source: LinkedIn Jobs. https://www.linkedin.com/jobs/view/artificial-intelligence-engineer-at-deephow-4437440707
2026-07-01: PharmaCloud launched for regulated manufacturing (gtm) . DeepHow introduced PharmaCloud, a GMP-compliant cloud environment for pharmaceutical and medical device manufacturers to standardize training, guide execution, and verify mission-critical work. ([deephow.com](https://deephow.com/blog/pharmacloud-standardize-train-and-verify-execution-in-a-gmp-compliant-environment)) Source: DeepHow blog. https://deephow.com/blog/pharmacloud-standardize-train-and-verify-execution-in-a-gmp-compliant-environment
2026-06-02: Foxconn launch with NVIDIA positioned Live SOP Verification as a physical-AI execution layer (gtm) . DeepHow announced a Foxconn deployment that uses NVIDIA Cosmos and Metropolis VSS for live SOP verification, aiming to detect procedural drift and improve first-pass yield and rework. ([deephow.com](https://deephow.com/blog/foxconn-boosts-production-throughput-with-deephow-live-sop-verification-powered-by-nvidia)) Source: DeepHow blog. https://deephow.com/blog/foxconn-boosts-production-throughput-with-deephow-live-sop-verification-powered-by-nvidia
2026-06-02: Live SOP Verification advanced through Foxconn deployment (product) . DeepHow announced a June 2, 2026 Foxconn deployment of Live SOP Verification powered by NVIDIA Cosmos and Metropolis VSS, describing real-time detection of procedural drift and execution verification on the factory floor. ([deephow.com](https://deephow.com/blog/foxconn-boosts-production-throughput-with-deephow-live-sop-verification-powered-by-nvidia?utm_source=openai)) Source: DeepHow Blog. https://deephow.com/blog/foxconn-boosts-production-throughput-with-deephow-live-sop-verification-powered-by-nvidia
2026-02-01: DeepHow expanded from training into execution verification and external sharing (gtm) . DeepHow launched Standard Work AI Verification and an external sharing add-on to connect training with proof of execution and extend content to customers, partners, and vendors. ([deephow.com](https://deephow.com/blog/standard-work-ai-verification-and-external-sharing-release)) Source: DeepHow blog. https://deephow.com/blog/standard-work-ai-verification-and-external-sharing-release
2025-05-28: Smart Compare and Smart Assist shown as newer factory-floor features (product) . At Automate 2025, DeepHow described Smart Compare for execution checking and Smart Assist for real-time answers from the training library, extending the platform from knowledge capture into in-flow guidance and verification. ([deephow.com](https://deephow.com/blog/deephow-at-automate-smarter-training-for-the-factory-floor?utm_source=openai)) Source: DeepHow Blog. https://deephow.com/blog/deephow-at-automate-smarter-training-for-the-factory-floor

Scale AI

Website: scale.com

Scale AI provides high-quality training data, evaluations, and applied AI tools, including a Physical AI Data Engine for robotics and other real-world AI systems.

Category: Robot Centric Demonstration & Teleoperation Data. Founded: 2016. Country: United States. Employees: 501-1,000 employees.

Landscape fit: Provides data infrastructure for collecting and evaluating robot demonstrations used to train physical AI systems.

Funding rounds

Strategic investment announced 2025-06-01 : 14300000000 USD ; investors: Meta Platforms, Inc. ; source: Reuters ; https://www.reuters.com/world/us/meta-finalizes-investment-scale-ai-valuing-startup-29-billion-2025-06-12/
Series F announced 2024-05-01 : 1000000000 USD ; investors: Accel, Y Combinator, Nat Friedman, Index Ventures, Founders Fund, Coatue, Thrive Capital, Spark Capital, NVIDIA, Tiger Global Management, Greenoaks, Wellington Management, Cisco Investments, DFJ Growth, Intel Capital, ServiceNow Ventures, AMD Ventures, WCM, Amazon, Elad Gil, Meta ; source: Business Wire ; https://www.businesswire.com/news/home/20240521674964/en/Scale-AI-Raises-%241-Billion-Series-F-to-Push-The-Frontier-of-AI-Data
Series E announced 2021-04-01 : 325000000 USD ; investors: Dragoneer, Greenoaks Capital, Tiger Global, Wellington Management, Durable Capital, Coatue, Index Ventures, Founders Fund, Y Combinator ; source: Scale AI ; https://scale.com/blog/series-e
Series D announced 2020-12-01 : 155000000 USD ; investors: Tiger Global ; source: Scale AI ; https://scale.com/blog/series-d
Series C announced 2019-08-01 : 100000000 USD ; investors: Founders Fund, Accel, Coatue Management, Index Ventures, Spark Capital, Thrive Capital, Kevin Systrom, Mike Krieger, Adam d’Angelo ; source: Scale AI ; https://scale.com/blog/series-c

Company timeline

2026-07-16: DOE Genesis Mission Consortium membership deepens execution role (gtm) . Scale joined the DOE Genesis Mission Consortium, moving from strategic partnership into active technical collaboration on model development, validation, data standards, HPC, and robotics. ([scale.com](https://scale.com/blog/scale-ai-joins-genesis-mission-consortium?utm_source=openai)) Source: Scale AI. https://scale.com/blog/scale-ai-joins-genesis-mission-consortium
2026-07-15: Mayo Clinic partnership validates healthcare deployment (gtm) . Scale announced work with Mayo Clinic to develop and deploy AI applications for clinical care using the Scale Generative AI Platform. ([scale.com](https://scale.com/blog/mayo-clinic-scale?utm_source=openai)) Source: Scale AI. https://scale.com/blog/mayo-clinic-scale
2026-07-09: Scale Labs posts multiple research and engineering openings (hr) . Scale Labs' jobs page listed multiple openings dated 2026-07-09, including AI infrastructure, ML systems, and research scientist roles, plus an ICML 2026 recruiting post.. Team size observed: 501-1,000 employees employees. Source: Scale Labs jobs. https://labs.scale.com/jobs
2026-05-20: GSA Schedule access opens a new government distribution channel (gtm) . Scale partnered with Carahsoft to place GenAI Platform, T&E, and professional services on GSA MAS for easier U.S. government procurement. ([scale.com](https://scale.com/blog/scale-ai-gsa-schedule-available?utm_source=openai)) Source: Scale AI. https://scale.com/blog/scale-ai-gsa-schedule-available
2026-05-01: DOE Genesis Mission MOU formalizes federal science partnership (gtm) . Scale signed an MOU with the DOE to support the Genesis Mission and collaborate on AI, advanced computing, and future joint projects. ([scale.com](https://scale.com/blog/scale-ai-doe-genesis-mission-mou?utm_source=openai)) Source: Scale AI. https://scale.com/blog/scale-ai-doe-genesis-mission-mou
2026-04-20: Scale acquired ICG Solutions for national security AI (product) . Scale said it acquired ICG Solutions and would integrate its real-time streaming data analytics with Scale’s agentic platform. Source: Scale AI. https://scale.com/blog/scale-acquires-icg-solutions

OpenGraph Labs

Website: opengraphlabs.com

OpenGraph Labs builds hardware and software for high-fidelity robotics data, including multimodal capture systems and tactile gloves for contact-rich robot learning.

Category: Robot Centric Demonstration & Teleoperation Data. Founded: 2025. Country: United States. Employees: 2-10 employees.

Landscape fit: Builds systems for capturing manipulation demonstrations and converting them into robot-learning datasets.

Company timeline

2026-07-19: 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/opengraph-labs
2026-07-01: SyncField multimodal data-capture system launched (gtm) . OpenGraph Labs announced SyncField, a multimodal data-capture system supporting arbitrary hardware stacks with automatic synchronization and fully time-aligned output. Source: OpenGraph Labs LinkedIn. https://www.linkedin.com/posts/julia-kim-62302a21b_robotics-worldmodels-physicalai-activity-7439049177847926785-oV2n
2026-07-01: Homepage sharpened around tactile ground-truth capture hardware (product) . The homepage now centers the product on a tactile capture glove, synchronized multimodal capture, and hardware-level control over the sensor stack. Source: OpenGraph Labs homepage. https://www.opengraphlabs.com/
2026-06-01: First public tactile manipulation sample released (gtm) . OpenGraph Labs published a public Ego-Tactile Manipulation sample on Hugging Face with synchronized egocentric video, dense two-hand tactile signals, and touch-grounded labels. Source: Hugging Face dataset. https://huggingface.co/datasets/OpenGraphLabs-Research/ego-tactile-manipulation
2026-06-01: Public physical-AI data hub launched with curated robotics datasets (gtm) . OpenGraph Labs launched a public DataHub for curated robotics, video, and simulation datasets, turning dataset curation into a front-door product surface. Source: OpenGraph Datahub. https://datahub.opengraphlabs.com/
2026-06-01: Multimodal capture configuration broadened across teleop and dexterous stacks (gtm) . The configurations page expanded the supported hardware matrix across human teleoperation, robot teleoperation, dexterous gloves, cameras, and object-state sensing. Source: OpenGraph Labs configurations. https://opengraphlabs.com/configurations

Lightwheel

Website: lightwheel.ai

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Series A. Founded: 2023. Country: United States. Employees: 201-500 employees.

Landscape fit: Combines teleoperation and real-to-simulation pipelines to create robot-native training data.

Funding rounds

Series A-V announced 2026-06-01 : 1000000000 CNY ; investors: Zhongguancun Science City Science and Technology Fund, Sichuan Development Sci-Tech Innovation Fund, Shandong Development Science & Technology Investment, Giant Interactive Group, Yusys Technologies, Boton Technology, Jiantou Huawen, 37 Interactive Entertainment, Liangtu Zhice, Senma Investment, Zhongke Industrial Investment, Undisclosed Investors ; source: Gasgoo ; https://autonews.gasgoo.com/articles/news/2069405877394952093
Series A++ / Series A+++ announced 2026-03-01 : 1000000000 CNY ; investors: CCB Sci-Tech, Guofang Innovation, Daohe Long-term Investment, Qingxin Capital, New Hope Group, Aux Electric, Capstone Capital, JIC Technology Investment, Shanghai Growth-FOF, Shanghai Daohe Long-term Investment Management ; source: MarketScreener UK ; https://uk.marketscreener.com/news/light-wheel-intelligence-technology-co-ltd-announced-that-it-has-received-cny-1-billion-in-fundin-ce7e5fd2de89fe22

Company timeline

2026-07-20: Lightwheel introduced SimReadyGen for agentic simulation generation (product) . Lightwheel announced SimReadyGen, described as agentic simulation generation for Physical AI. ([lightwheel.ai](https://www.lightwheel.ai/blogs?utm_source=openai)) Source: Blogs index. https://www.lightwheel.ai/blogs
2026-07-02: Geely deployment case showed Real2Sim2Real in production (gtm) . Lightwheel published a deployment story showing Geely’s humanoid robots operating on a production line using Lightwheel’s Real2Sim2Real loop, with week-scale iteration and an order-of-magnitude cost reduction. ([lightwheel.ai](https://lightwheel.ai/media/geely-humanoid-robots?utm_source=openai)) Source: Lightwheel / Geely case study. https://www.lightwheel.ai/media/geely-humanoid-robots
2026-07-01: MANUS partnership expanded the human-data capture stack (gtm) . Lightwheel and MANUS announced a strategic partnership to combine high-precision hand capture with Lightwheel’s human-data infrastructure for dexterous manipulation and joint go-to-market. ([lightwheel.ai](https://lightwheel.ai/media/lightwheel-manus-partnership?utm_source=openai)) Source: Lightwheel / MANUS. https://www.lightwheel.ai/media/lightwheel-manus-partnership
2026-07-01: LinkedIn company page shows 201-500 employees (hr) . Lightwheel’s LinkedIn company page lists company size as 201-500 employees.. Team size observed: 201-500 employees employees. Source: LinkedIn. https://www.linkedin.com/company/lightwheel2026
2026-06-01: Multi-role hiring push appears on careers and Ashby job board (hr) . Lightwheel’s careers page and job board show active openings for Go-To-Market / Technical Sales Manager, Robotics Solution Architect, Developer Advocate / Research Community Intern, Global Field Marketing Manager, and Founding HR.. Team size observed: 201-500 employees employees. Source: Lightwheel careers page and Ashby job posts. https://www.lightwheel.ai/careers
2026-06-01: Series A-V (funding) . Amount: 1000000000 CNY. Investors: Zhongguancun Science City Science and Technology Fund, Sichuan Development Sci-Tech Innovation Fund, Shandong Development Science & Technology Investment, Giant Interactive Group, Yusys Technologies, Boton Technology, Jiantou Huawen, 37 Interactive Entertainment, Liangtu Zhice, Senma Investment, Zhongke Industrial Investment, Undisclosed Investors Source: Gasgoo. https://autonews.gasgoo.com/articles/news/2069405877394952093

RLWRLD Inc.

Website: rlwrld.ai

RLWRLD builds manipulation foundation models and a robotics intelligence layer that train industrial robots to handle dexterous, human-like tasks using real-world operational data.

Category: Robot Centric Demonstration & Teleoperation Data. Founded: 2023. Country: Japan. Employees: 51-200 employees.

Landscape fit: Captures skilled human motion and converts it into training data for robotic manipulation.

Funding rounds

Seed 2 announced 2026-02-01 : 26000000 USD ; investors: Headline Asia, Z Venture Capital, CJ Logistics, Kakao Investment, Lotte Ventures, Hanwha Asset Management, Mirae Asset – Emart Investment Fund I, Hyosung Ventures, Smilegate Investment, T Investment, Hashed Ventures ; source: GlobeNewswire ; https://www.globenewswire.com/news-release/2026/02/26/3245146/0/en/rlwrld-raises-26m-seed-2-bringing-total-funding-to-41m-to-scale-industrial-robotics-ai.html
Seed announced 2025-04-01 : 15000000 USD ; investors: LG Electronics, SK Telecom, KDDI, ANA Holdings, Mitsui Chemicals, Shimadzu Corporation, Hashed, Mirae Asset, Global Brain ; source: PR Newswire ; https://www.prnewswire.com/news-releases/rlwrld-raises-15m-to-build-gpt-for-robots-backed-by-asias-industrial-giants-302427574.html

Company timeline

2026-07-01: RLWRLD LinkedIn company size visible as 51-200 employees (hr) . RLWRLD’s LinkedIn company page lists the company size as 51-200 employees.. Team size observed: 51-200 employees employees. Source: LinkedIn. https://www.linkedin.com/company/rlwrld/
2026-07-01: RLWRLD careers page shows broad active hiring cluster (hr) . RLWRLD’s careers page currently lists multiple open roles across Tech/Product, Business, Strategy, Robotics Deployment, Operation, and PR, including research engineers, software engineers, deployment roles, business development, and a PR & Social Media Manager.. Team size observed: 51-200 employees employees. Source: RLWRLD Careers. https://www.rlwrld.ai/en/careers
2026-06-29: All Hands Up! robot-hand analysis platform (product) . RLWRLD announced All Hands Up!, a robot-hand analysis platform that provides technical reporting and visualization around dexterous robot hand design tradeoffs. Source: RLWRLD Japanese News. https://www.rlwrld.ai/jp/news/79
2026-06-19: Expanded into hotel robotics transformation planning with Lotte (gtm) . RLWRLD was named as a partner to Lotte Hotels & Resorts in a long-term robot transformation plan covering hotel-specific workflows and phased humanoid deployment. ([rlwrld.ai](https://www.rlwrld.ai/en/news/63?utm_source=openai)) Source: RLWRLD. https://www.rlwrld.ai/en/news/63
2026-06-19: Partnered with NVIDIA on dexterity benchmarks and data standards (gtm) . RLWRLD announced a collaboration with NVIDIA to develop DexBench, a humanoid dexterity benchmark, plus a shared data standard and deeper Isaac Lab integration. ([rlwrld.ai](https://www.rlwrld.ai/en/news/58?utm_source=openai)) Source: RLWRLD. https://www.rlwrld.ai/en/news/58
2026-06-19: RLDX-1 model launch details and benchmark results (product) . RLWRLD reported that RLDX-1 was officially unveiled on May 7 and showcased at NVIDIA GTC Taipei 2026 with benchmark results across tabletop, kitchen manipulation, and coffee-pouring tasks. Source: RLWRLD News. https://www.rlwrld.ai/en/news/59

Shift

Website: joinshift.com

Shift is a robotics training-data company that pays people to record everyday household and trade tasks, then turns those recordings into anonymized datasets for AI and robotics customers.

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Seed. Founded: 2025. Country: United States.

Landscape fit: Captures physical task demonstrations through cameras and sensor-equipped wearables for robotics training.

Funding rounds

Seed announced 2026-06-01 : 10200000 USD ; investors: Ace Capital Partners, Aurelius Capital Management, Corner Ventures, Arkin Capital, STEP World, Fresh Fund ; source: GlobeNewswire ; https://www.globenewswire.com/news-release/2026/06/03/3305884/0/en/shifters-raises-10-2-million-seed-round-led-by-ace-capital-partners-to-advance-ai-native-ground-robotics.html

Company timeline

2026-07-13: Privacy policy added anonymized robotics datasets and API subscriptions (product) . Shift’s privacy policy now describes a platform for managing and distributing anonymized robotics datasets, including commercial subscriptions through a platform API for AI and robotics development. ([joinshift.us](https://www.joinshift.us/privacy?utm_source=openai)) Source: Shift Privacy Policy. https://www.joinshift.us/privacy
2026-07-01: Self-serve business onboarding page for Shift for Business (product) . Shift added a business registration flow labeled Shift for Business and a request-a-contract entry point for organizations. ([app.joinshift.us](https://app.joinshift.us/apply/b2b?utm_source=openai)) Source: Shift for Business. https://app.joinshift.us/apply/b2b
2026-07-01: Formalized operator workflow with headset, app, and weekly payouts (product) . Shift’s live site described a five-step operator flow: apply, purchase a headset, download the app, record tasks, and get paid weekly for accepted uploads. ([joinshift.us](https://www.joinshift.us/)) Source: Shift homepage. https://www.joinshift.us/
2026-07-01: Active hiring cluster appears on LinkedIn (hr) . LinkedIn’s Shift jobs page showed 3 open roles: Internship Performance Marketing in Munich, Sales & Operations Representative – Sicilia in Sicily, and Video Annotation Lead - Labeling (Robotics) in India. Source: LinkedIn. https://www.linkedin.com/jobs/shift-jobs-worldwide?f_C=119124010&trk=top-card_top-card-primary-button-top-card-primary-cta
2026-06-20: Released operator terms for egocentric video capture and task platform use (product) . Shift published operator terms covering onboarding, recording, segmentation, annotation, quality control, payout workflows, and head-mounted egocentric capture using Shift-approved equipment. ([joinshift.us](https://www.joinshift.us/terms)) Source: Shift Terms and Conditions. https://www.joinshift.us/terms
2026-06-01: NYC cleaning launch to collect first-person robotics data (product) . Shift launched by offering free apartment cleaning in New York City in exchange for recording the work with a company device, framing the service as a way to fund robotics training data collection. ([linkedin.com](https://www.linkedin.com/company/joinshift/)) Source: LinkedIn. https://www.linkedin.com/company/joinshift/

XDOF, Inc.

Website: xdof.ai

XDOF builds data pipelines, collection tools, annotation systems, and teleoperation-driven datasets to help robotics and frontier AI labs train robots for physical-world tasks.

Category: Robot Centric Demonstration & Teleoperation Data. Founded: 2024. Country: United States. Employees: 11-50 employees.

Landscape fit: Builds systems for collecting high-quality robot demonstrations and action trajectories at scale.

Funding rounds

Series A announced 2026-06-01 : 70000000 USD ; investors: Thrive Capital, Spark Capital, Andreessen Horowitz, Lux Capital, WndrCo ; source: Cooley ; https://www.cooley.com/news/coverage/2026/2026-06-17-xdof-emerges-from-stealth-with-%2470-million-raise
Seed announced 2024-10-01 : 7840000 USD ; investors: Samsung Next, WndrCo ; source: Forge Global ; https://forgeglobal.com/x-dof_ipo/

Company timeline

2026-07-21: XDOF careers page shows 18 open roles (hr) . The company careers page currently lists 18 open positions across robotics, software engineering, mechanical engineering, product, operations, GTM, and general & administrative functions.. Team size observed: 11-50 employees employees. Source: XDOF Careers. https://www.xdof.ai/careers
2026-07-17: Founder announces XDOF is coming out of stealth (hr) . Co-founder and CEO Philipp Wu publicly announced that XDOF is coming out of stealth and said the company had already been working behind the scenes with major robotics labs and companies.. Team size observed: 11-50 employees employees. Source: LinkedIn. https://www.linkedin.com/in/wuphilipp
2026-06-30: Positioned data quality and annotation tooling as a product surface (gtm) . XDOF published a research post about teaching robots to distinguish productive moments from dead time, emphasizing tooling around higher-quality robot data. Source: XDOF Blog. https://www.xdof.ai/blog
2026-06-30: Launched WARP for self-supervised data quality weighting (product) . XDOF published WARP-RM, a method that uses warped playback to create pseudo-labels and weight robot demonstrations by task progress. Source: XDOF Blog. https://www.xdof.ai/blog/warp-rm
2026-06-23: Used research publication to validate robot-data quality and self-improvement loop (gtm) . XDOF published SARM2 research framing reward modeling as a way to grade tasks and improve policies from raw demonstrations. Source: XDOF Blog. https://www.xdof.ai/blog
2026-06-23: Introduced SARM2 reward-modeling for robot data (product) . XDOF published SARM2, a reward model that grades manipulation tasks and turns raw demonstrations into improving policies. Source: XDOF Blog. https://www.xdof.ai/blog/sarm2

Mecka AI

Website: mecka.ai

Mecka AI builds the data and deployment layer for physical AI, collecting and structuring real-world human activity data so robotics and embodied-AI teams can train and deploy systems more reliably.

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Series A. Founded: 2024. Country: Canada. Employees: 11-50 employees.

Landscape fit: Provides infrastructure for collecting and evaluating human and robot interaction data for physical AI.

Funding rounds

Follow-on investment announced 2026-06-01 : 35000000 USD ; investors: Framework Ventures, Menlo Ventures, SV Angel, Kindred Ventures, Ted Xiao ; source: Fortune ; https://fortune.com/2026/06/01/mecka-ai-series-a-60-million-robotics-data-training/
Series A announced 2026-06-01 : 25000000 USD ; investors: Framework Ventures, Menlo Ventures, SV Angel, Kindred Ventures, Ted Xiao ; source: Fortune ; https://fortune.com/2026/06/01/mecka-ai-series-a-60-million-robotics-data-training/
Seed announced 2025-08-01 : 8000000 USD ; investors: Neo, SV Angel, A-Star Capital, Cade, Timeless, Offline VC, Mechanism Capital ; source: Silicon Legal Strategy ; https://www.siliconlegal.com/sls-client-neo-leads-mecka-ais-8mm-seed-round/

Company timeline

2026-07-21: LinkedIn employee count indicates early scaling team size (hr) . Mecka’s LinkedIn company page shows a company size of 11-50 employees.. Team size observed: 11-50 employees employees. Source: LinkedIn. https://www.linkedin.com/company/mecka-ai
2026-07-21: Hiring page shows broad multi-function recruiting push (hr) . Mecka’s careers page lists 36 open roles across 8 teams, including research, platform engineering, hardware and robotics, field engineering, operations, account management/sales, finance/legal, and marketing.. Team size observed: 11-50 employees employees. Source: Mecka careers page. https://www.mecka.ai/careers
2026-07-01: Homepage repositioned Mecka as data plus deployment layer (product) . The public homepage now frames Mecka as the data and deployment layer for physical AI and highlights dataset access, data tools, inference APIs, and enterprise evaluation/deployment support. ([mecka.ai](https://www.mecka.ai/?utm_source=openai)) Source: Mecka homepage. https://www.mecka.ai/
2026-06-01: Partner web app surfaced dataset review and episode workflows (product) . Mecka’s partner web app exposed operational surfaces for submissions, processed episodes, labels, tasks, users, and groups, indicating a production workflow for reviewing and managing collected data. ([points.mecka.ai](https://points.mecka.ai/dashboard/processed_episodes?utm_source=openai)) Source: Mecka partner/dashboard pages. https://points.mecka.ai/dashboard/processed_episodes
2026-06-01: Mecka App documented for iPhone-based data capture (product) . EgoVerse documentation added a Mecka App workflow for iPhone-based capture, including sign-in, task recording, review, and upload steps, and the homepage says the mobile app is now available on iOS. ([egoverse.ai](https://egoverse.ai/hardware/mecka-app.html?utm_source=openai)) Source: EgoVerse hardware docs / Mecka homepage. https://egoverse.ai/hardware/mecka-app.html
2026-06-01: Follow-on investment (funding) . Amount: 35000000 USD. Investors: Framework Ventures, Menlo Ventures, SV Angel, Kindred Ventures, Ted Xiao Source: Fortune. https://fortune.com/2026/06/01/mecka-ai-series-a-60-million-robotics-data-training/

Sensei Robotics

Website: senseirobotics.com

Sensei Robotics builds robotic training data infrastructure, connecting robotics companies to human operators who collect in-the-wild demonstration data for robot learning.

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Pre Seed. Founded: 2024. Country: United States. Employees: 2-10 employees.

Landscape fit: It fits this category because its core product is robot-native demonstration data gathered via human teleoperation of robots, which is exactly the kind of robot-centric training data used to train robotics models. The company says it provides 'robotic training data, at scale' and connects robotics companies to human operators collecting demonstrations, while YC describes its platform as collecting human-demonstration data and a marketplace for robotics training data. ([senseirobotics.com](https://senseirobotics.com/))

Funding rounds

Pre-Seed announced 2024-10-01 : 500000 USD ; investors: Y Combinator ; source: LinkedIn ; https://www.linkedin.com/company/sensei-robotics-yc-s24/

Company timeline

2026-07-01: Beta access page added for platform and hardware (product) . The website now invites prospective users to join a beta for early access to Sensei’s platform and hardware. Source: Sensei Robotics website. https://senseirobotics.com/beta
2026-07-01: Public LinkedIn page shows a 2-10 employee company size (hr) . Sensei Robotics’ LinkedIn company page shows company size as 2-10 employees and lists a small employee roster.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/company/sensei-robotics-yc-s24
2026-07-01: Website adds two-sided commercial funnels for buyers and operators (gtm) . The live site now splits GTM into a buyer funnel for requesting data and an operator funnel for joining the beta and teaching robots to be human. Source: Sensei Robotics. https://senseirobotics.com/
2026-07-01: Current site added dedicated request-data flow (product) . The website now includes a structured request-data intake flow that asks for data type, volume, deadline, and required hardware. Source: Sensei Robotics website. https://senseirobotics.com/request-data
2024-10-01: Pre-Seed (funding) . Amount: 500000 USD. Investors: Y Combinator Source: LinkedIn. https://www.linkedin.com/company/sensei-robotics-yc-s24/
2024-08-01: Sensei launched as a robotics training-data platform (product) . Sensei introduced itself as a combined hardware and software platform for collecting human demonstration data at scale, supported by a network of paid operators. Source: Y Combinator Launch YC. https://www.ycombinator.com/launches/Ljf-sensei-robotics-training-data-at-scale

Humanola

Website: humanola.com

Humanola is a robotics data infrastructure company that provides a teleoperation platform and SDK to capture, sync, and convert robot demonstrations into training-ready datasets for embodied AI.

Category: Robot Centric Demonstration & Teleoperation Data. Funding stage: Seed. Founded: 2025. Country: United States. Employees: 2-10 employees.

Landscape fit: Records teleoperated robot demonstrations and converts them into training-ready robotics datasets.

Funding rounds

Seed announced 2025-12-01 : 2500000 USD ; investors: Link Ventures, Bad Ideas Fund, Georgia Tech Fund, Akmal Payziyev ; source: Dealroom.co ; https://app.dealroom.co/companies/humanola

Company timeline

2026-07-01: Sample dataset delivery motion (gtm) . Humanola published a branded sample dataset on Hugging Face, showing real egocentric teleoperation episodes and a request-a-sample style evaluation flow. ([huggingface.co](https://huggingface.co/datasets/humanola-inc/egocentric-sample)) Source: Hugging Face dataset by humanola-inc. https://huggingface.co/datasets/humanola-inc/egocentric-sample
2026-07-01: Productized into platform, pricing, and enterprise integration (gtm) . Humanola’s current site now presents a three-part platform—capture, store, process—with a live SDK, dataset licensing, a Free/Premium/Enterprise price structure, and custom integration for enterprise deployments. ([humanola.com](https://humanola.com/)) Source: Humanola website and docs. https://humanola.com/
2026-07-01: Current small-team size visible on LinkedIn (hr) . Humanola's LinkedIn page shows a 2-10 employee company size and 6 visible employees.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/company/humanola
2026-07-01: Platform expanded into teleoperation-to-dataset infrastructure (product) . Humanola’s main site shifted to a broader data infrastructure offering with a Python SDK, live Capture and Store, Process in beta, and License in early access. Source: Humanola homepage. https://humanola.com/
2026-06-01: Founder featured in public update about Humanola's physical AI mission (hr) . Humanola posted that founder Sanjar Atamuradov joined Jean Lee to discuss the robotics data bottleneck and the company’s teleoperation logging, multimodal sync, and training-ready dataset pipeline.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/posts/humanola_our-founder-sanjar-atamuradov-joined-jean-activity-7473454948299603968-Bksq
2026-04-01: AgiBot cooperation discussion (gtm) . A LinkedIn post from Vladimir Norov describes a productive discussion with AgiBot about cooperation with Humanola on embodied AI and humanoid robotics for industrial and service environments. ([linkedin.com](https://www.linkedin.com/posts/vladimirnorov_robotics-humanoidrobots-embodiedai-activity-7410658143933956097-2Szs?utm_source=openai)) Source: LinkedIn post by Vladimir Norov. https://www.linkedin.com/posts/vladimirnorov_robotics-humanoidrobots-embodiedai-activity-7410658143933956097-2Szs

Aktoria Robotics

Website: aktoria.com

Aktoria Robotics builds remote robot teleoperation and data-collection software that lets humans operate robots anywhere and capture robot-native demonstrations for training and fleet support.

Category: Robot Centric Demonstration & Teleoperation Data. Founded: 2026. Country: United States. Employees: 2.

Landscape fit: It fits this category because its core product is robot-native teleoperation and demonstration capture, which directly creates training data and operational trajectories for robotics models.

Funding rounds

seed announced 2026-01-01 : 500000 USD ; investors: Y Combinator ; source: Y Combinator ; https://www.ycombinator.com/companies/aktoria-robotics

Company timeline

2026-07-01: Founding team publicly listed on YC company page (hr) . YC’s company profile lists Aktoria Robotics as active in Summer 2026 with two active founders and a team size of 2.. Team size observed: 2 employees. Source: Y Combinator. https://www.ycombinator.com/companies/aktoria-robotics
2026-07-01: YC-backed launch with remote-operations positioning (gtm) . Aktoria Robotics surfaced publicly as a YC Summer 2026 company and launched a website framing the product as "Human as API for Robots" with teleoperation, remote data collection, fleet monitoring, and a book-a-call CTA. Source: Y Combinator / Aktoria Robotics. https://www.ycombinator.com/companies/aktoria-robotics
2026-07-01: Public launch of remote teleoperation and data-collection platform (product) . Aktoria Robotics publicly presented a product focused on remote robot teleoperation, data collection on customer robots, fleet monitoring and intervention, and a train→deploy→monitor workflow. Source: Aktoria Robotics homepage. https://aktoria.com/
2026-01-01: seed (funding) . Amount: 500000 USD. Investors: Y Combinator Source: Y Combinator. https://www.ycombinator.com/companies/aktoria-robotics

Training Data Curation & Verification

Companies that evaluate robot demonstrations and rollouts so raw recordings can become reliable training data for robotics models.

Instance

Website: instancelabs.ai

Instance builds automated evaluation software for robot policies, starting with a verifier that judges whether robot rollouts succeeded and breaks tasks into subtasks from video data.

Category: Training Data Curation & Verification. Founded: 2026. Country: United States. Employees: 2-10 employees.

Landscape fit: It fits the category because it evaluates robot demonstrations and rollouts to turn raw episodes into structured, reliable training/evaluation data for robotics models.

Company timeline

2026-07-01: Main site repositioned around ground truth for robot learning (gtm) . Instance's main site shifted to a sharper positioning as the verification layer for robot learning, emphasizing trusted data, verdicts from real episodes, and a robotics-specific workflow. Source: Instance homepage. https://www.instancelabs.ai/
2026-07-01: Public verifier demo launched (gtm) . Instance launched a public rollout-verification demo that judges task attempts, splits multi-attempt videos, exposes HTTP API endpoints, and highlights benchmark and latency advantages. Source: Instance demo site. https://demo.instancelabs.ai/
2026-04-01: Early-access waitlist opened (gtm) . Instance introduced a waitlist for early access to the product, signaling an initial founder-led demand capture motion before broad public availability. Source: Instance waitlist page. https://demo.tryinstance.app/waitlist
2026-02-01: Instance launched by founders Claire Mao and Lucy Cai (hr) . Claire Mao publicly introduced Instance and described the company as an AI data analyst/robot-learning product being built with Lucy Cai.. Team size observed: 2-10 employees employees. Source: LinkedIn. https://www.linkedin.com/in/clairemao-

Shotwell

Website: shotwell.ai

Category: Training Data Curation & Verification. Founded: 2026. Country: United States. Employees: 2-10 employees.

Landscape fit: It fits this category because its core product evaluates robot demonstrations and rollouts so raw recordings can be turned into reliable training and verification data for robotics models. ([ycombinator.com](https://www.ycombinator.com/companies/shotwellai?utm_source=openai))

Funding rounds

seed announced 2026-01-01 : 500000 USD ; investors: Y Combinator ; source: Y Combinator ; https://www.ycombinator.com/companies/shotwellai

Company timeline

2026-07-01: Early credibility through named trust signals (gtm) . The homepage added a "Trusted by" section naming Parametric, indicating the first visible customer-proof motion on the public site. ([shotwell.ai](https://shotwell.ai/)) Source: Shotwell website. https://shotwell.ai/
2026-07-01: Demo-led workflow proof for annotation and review (gtm) . Shotwell exposed a live demo showing episode review, JSON export, annotation controls, and failure/pass review flows, reinforcing a technical buyer motion around workflow proof rather than generic marketing. ([demo.shotwell.ai](https://demo.shotwell.ai/)) Source: Shotwell demo site. https://demo.shotwell.ai/
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/shotwell-ai
2026-07-01: Public launch of robotics observability positioning (gtm) . Shotwell publicly introduced itself as "the observability layer for robotics," with messaging focused on fast, accurate, dense annotations for robotics training data and failure detection in both training videos and live deployments. ([shotwell.ai](https://shotwell.ai/)) Source: Shotwell website / YC company page. https://shotwell.ai/
2026-01-01: seed (funding) . Amount: 500000 USD. Investors: Y Combinator Source: Y Combinator. https://www.ycombinator.com/companies/shotwellai

Investors

Top-tier and active investors

Top-tier investors who invested

Did not invest

Other investors

Funding & Exits

Jul 2026

6thSense

Pre-Seed

1B KRW • Investors: Krew Capital

VentureSquare

Jun 2026

Praxis AI

Seed

$125K • Investors: Y Combinator

Dealroom

Mecka AI

Acquisition

Acquirer: Mecka AI • Status: Announced

BetaKit

Mecka AI

Follow-on investment

$35M • Investors: Framework Ventures, Menlo Ventures, SV Angel, Kindred Ventures, Ted Xiao

Fortune

Mecka AI

Series A

$25M • Investors: Framework Ventures, Menlo Ventures, SV Angel, Kindred Ventures, Ted Xiao

Fortune

XDOF, Inc.

Series A

$70M • Investors: Thrive Capital, Spark Capital, Andreessen Horowitz, Lux Capital, WndrCo

Cooley

Shift

Seed

$10.2M • Investors: Ace Capital Partners, Aurelius Capital Management, Corner Ventures, Arkin Capital, STEP World, Fresh Fund

GlobeNewswire

Lightwheel

Series A-V

1B CNY • Investors: Zhongguancun Science City Science and Technology Fund, Sichuan Development Sci-Tech Innovation Fund, Shandong Development Science & Technology Investment, Giant Interactive Group, Yusys Technologies, Boton Technology, Jiantou Huawen, 37 Interactive Entertainment, Liangtu Zhice, Senma Investment, Zhongke Industrial Investment, Undisclosed Investors

Gasgoo

May 2026

Human Archive

Seed

$8.2M • Investors: Wing Venture Capital, NVP Capital, Y Combinator, angel investors from OpenAI, angel investors from Nvidia, angel investors from Google, angel investors from Mercor, angel investors from AfterQuery, angel investors from BAIR, angel investors from SAIL, Brad Boa, angel investors from Meta

Human Archive official website

Luel

Seed

$31.2M • Investors: Lightspeed Venture Partners, General Catalyst, SV Angel

Lightspeed Venture Partners

Mar 2026

Lightwheel

Series A++ / Series A+++

1B CNY • Investors: CCB Sci-Tech, Guofang Innovation, Daohe Long-term Investment, Qingxin Capital, New Hope Group, Aux Electric, Capstone Capital, JIC Technology Investment, Shanghai Growth-FOF, Shanghai Daohe Long-term Investment Management

MarketScreener UK

Feb 2026

RLWRLD Inc.

Seed 2

$26M • Investors: Headline Asia, Z Venture Capital, CJ Logistics, Kakao Investment, Lotte Ventures, Hanwha Asset Management, Mirae Asset – Emart Investment Fund I, Hyosung Ventures, Smilegate Investment, T Investment, Hashed Ventures

GlobeNewswire

Jan 2026

Shotwell

Seed

$500K • Investors: Y Combinator

Y Combinator

Aktoria Robotics

Seed

$500K • Investors: Y Combinator

Y Combinator

Luel

Seed

$125K • Investors: Y Combinator

Dealroom

Dec 2025

Humanola

Seed

$2.5M • Investors: Link Ventures, Bad Ideas Fund, Georgia Tech Fund, Akmal Payziyev

Dealroom.co

Nov 2025

Cortex AI

Seed VC

$6M • Investors: 500 Global, Gokul Rajaram, Golden Gate Ventures, Kaz Nejatian, Liquid 2 Ventures, Matrix Partners, Pioneer Fund, Spot VC, Y Combinator

CB Insights

Sep 2025

Hub

Seed

$1.7M • Investors: SwissBorg

Business Wire

Build AI

Seed

$5M • Investors: Abstract Ventures, Pear VC, HF0, ZFellows, Alex Botez

StartupHub.ai

Aug 2025

Mecka AI

Seed

$8M • Investors: Neo, SV Angel, A-Star Capital, Cade, Timeless, Offline VC, Mechanism Capital

Silicon Legal Strategy

Jul 2025

Claru (Reka AI, Inc.)

Series B

$110M • Investors: NVIDIA, Snowflake

Reka

Jun 2025

Scale AI

Strategic investment

$14.3B • Investors: Meta Platforms, Inc.

Reuters

May 2025

DeepHow

Convertible Note

$5M

CB Insights

Apr 2025

RLWRLD Inc.

Seed

$15M • Investors: LG Electronics, SK Telecom, KDDI, ANA Holdings, Mitsui Chemicals, Shimadzu Corporation, Hashed, Mirae Asset, Global Brain

PR Newswire

Oct 2024

Sensei Robotics

Pre-Seed

$500K • Investors: Y Combinator

XDOF, Inc.

Seed

$7.8M • Investors: Samsung Next, WndrCo

Forge Global

Jul 2024

Instawork

Acquisition

Acquirer: Instawork • Status: Announced

Instawork

May 2024

Scale AI

Growth

$1B • Investors: Accel, Y Combinator, Nat Friedman, Index Ventures, Founders Fund, Coatue, Thrive Capital, Spark Capital, NVIDIA, Tiger Global Management, Greenoaks, Wellington Management, Cisco Investments, DFJ Growth, Intel Capital, ServiceNow Ventures, AMD Ventures, WCM, Amazon, Elad Gil, Meta

Business Wire

Jun 2023

Claru (Reka AI, Inc.)

Series A

$60M • Investors: DST Global Partners, Radical Ventures, Snowflake Ventures, Nat Friedman

Reka

May 2023

Instawork

Growth

$60M • Investors: TCV, 9Yards Capital, Larry Fitzgerald Jr., Benchmark, Spark Capital, Craft Ventures, Greylock, Four Rivers Group, GV

Instawork

Apr 2023

DeepHow

Series A

$14M • Investors: Owl Ventures, LG Technology Ventures, Sierra Ventures, Qualcomm Ventures, Osage Venture Partners, Foothill Ventures

PR Newswire

Sep 2022

Claru (Reka AI, Inc.)

Series Seed

$7.4M

Forge Global

Sep 2021

DeepHow

Pre-Series A

$9M • Investors: Sierra Ventures, Osage Venture Partners, Qualcomm Ventures LLC, Foothill Ventures

PR Newswire

Jul 2021

Instawork

Series C

$60M • Investors: Craft Ventures, Greylock, Corner Ventures, Four River Group, WndrCo, Tilman Fertitta, Benchmark, Spark Capital, GV, Burst Capital, SV Angel

PR Newswire

Apr 2021

Scale AI

Growth

$325M • Investors: Dragoneer, Greenoaks Capital, Tiger Global, Wellington Management, Durable Capital, Coatue, Index Ventures, Founders Fund, Y Combinator

Scale AI

Dec 2020

Scale AI

Growth

$155M • Investors: Tiger Global

Scale AI

Aug 2019

Scale AI

Series C

$100M • Investors: Founders Fund, Accel, Coatue Management, Index Ventures, Spark Capital, Thrive Capital, Kevin Systrom, Mike Krieger, Adam d’Angelo

Scale AI

May 2019

Instawork

Series B

$18M • Investors: Spark Capital, GV, Burst Capital, Benchmark, Y Combinator, Tuesday Capital, SV Angel

Instawork

Aug 2018

Scale AI

Series B

$18M • Investors: Index Ventures, Accel, Y Combinator, Drew Houston, Justin Kan

Scale AI

Jul 2018

DeepHow

Seed

$120K • Investors: Fontinalis Partners, Right Side Capital Management

Dealroom.co

Feb 2018

Instawork

Series A

$8.2M • Investors: Benchmark, Y Combinator, Tuesday Capital, Liquid 2 Ventures, SV Angel

Global Venturing

Jul 2017

Scale AI

Series A

$4.5M • Investors: Accel

Scale AI

On mobile, only the 10 most recent funding and exit groups are shown.