Map the next
intelligence.^↗

01Abundant

RL environments and datasets for learning through simulated experience.

02AfterQuery

Capturing how experts think, reason, and work as training data for AI.

03AIChamp

Video-derived training data and evaluations for physical AI.

04Akhara

Verification for agents using enterprise knowledge and expert judgment.

05Andon Labs

Benchmarks and experiments for autonomous agents running businesses.

06Andromede

Programmatically generated RL tasks, environments, and verifiers.

07Anthromind

Expert human data for RLHF, fine-tuning, and evaluation.

08Appen

Human training data, annotation, and agentic RL environments.

09Aptura

Expert datasets and RL environments for regulated industries.

10ARIMLABS

Verifiable RL environments for cybersecurity and technical work.

11BenchFlow

Environments, benchmarks, and runtimes for training and testing agents.

12Bespoke Labs

Company-scale environments for training agents on complex, long-horizon work.

13Chakra Labs

Software replicas and trajectory datasets for computer-use agents.

14Collinear

Enterprise RL environments, verifiers, and training data.

15Cua

Computer-use sandboxes and tools for agent training and evaluation.

16Datacurve

Research and expert data collection to expand what AI can do.

17DataVendor

A marketplace connecting private data, code, and expertise with AI buyers.

18Daytona

Isolated, programmable infrastructure for running agent code.

19Deeptune

Software-workflow training environments; joining Mercor.

20Defined.ai

A marketplace for licensed, multimodal AI training datasets.

21Dissei

Financial-workflow data, trajectories, and verifiable RL tasks.

22dmodel

RL environments for interpretability and alignment research agents.

23E2B

Cloud sandboxes for coding, computer use, and RL rollouts.

24EdotEnv

Financial-market environments for training research agents.

25Emulated

Post-training data and environments for coding and AI research.

26Exabite

Text, document, and video datasets for model training.

27Fleet

Simulated worlds and real-world challenges for more capable AI agents.

28General Reasoning

Long-horizon reinforcement learning and open agent research.

29Good Start Labs

Game environments and gameplay data for learning agents.

30Halluminate

RL environments for finance and professional-services workflows.

31Handshake AI

Human experts, training data, and environments for frontier models.

32Hillclimb

Training data and environments for self-improving research agents.

33HUD

Infrastructure for building, evaluating, and training agents in RL environments.

34Human Native

AI data licensing marketplace joining Cloudflare.

35Huzzle Labs

Long-horizon environments for computer-use and enterprise agents.

36Idler

RL environments built around production engineering work.

37iMerit

Specialist annotation, reasoning data, and human feedback.

38Incalmo

Cybersecurity environments for reinforcement learning and self-play.

39Innodata

Data engineering, annotation, and expert services for generative AI.

40Invisible Technologies

Human expertise and data operations for training AI models.

41Labelbox

Expert data, RL environments, and robotics training supply.

42Latch

Scientific workflows and benchmarks for biological AI agents.

43Markov

Computer-use demonstrations and design-software environments.

44Matrices

Computer-use environments for training multimodal agents.

45Mechanize

Software engineering environments that push frontier coding agents further.

46Mercor

Expert-created datasets, benchmarks, and environments for frontier AI.

47Metaphi

Enterprise coding environments, datasets, and evaluations.

48micro1

Expert training data, simulated work environments, and robotics data.

49Modal

Compute and sandboxes for RL training and parallel agent rollouts.

50Morph

Branching execution environments and scalable infrastructure for agents.

51Normal

Data and evaluations for AI systems that design hardware.

52Ooak Data

Enterprise data and digital twins for training AI agents.

53Pareto

Human expertise and feedback for training and evaluating AI.

54Parsewave

Human-authored engineering tasks and traces for model training.

55Patronus AI

Simulated worlds and evaluations for training AI agents.

56Plato

Simulated browser environments for agent training and evaluation.

57pre.dev

Verified software-engineering tasks for coding-agent training.

58Preference Model

RL environments for automated machine-learning research.

59Prime Intellect

An integrated stack for RL environments, model training, and compute.

60Prolific

Human participants and feedback for AI training and evaluation.

61Proximal

Software-engineering training data and long-horizon coding benchmarks.

62ReasonCore

Expert-curated training data across technical and professional domains.

63Refresh

Software environments for coding and computer-use agent training.

64Rise Data Labs

Human expert data for model training and evaluation.

65Runloop

Coding sandboxes, benchmarks, and infrastructure for agent training.

66Sama

Image, video, and other annotated data for AI development.

67Scale AI

Human-generated training data and evaluations for AI development.

68Snorkel AI

Expert-built datasets, environments, and benchmarks for AI agents.

69Surge AI

Human intelligence and expert data for training the next generation of AI.

70Symbal

Multimodal post-training data and expert evaluations.

71Taste Labs

Human design judgment for training and evaluating AI.

72Theta

Workspace environments and expert data for long-horizon agents.

73Toloka

Expert training data and environments for agents and language models.

74Trajectory Labs

Safety evaluations, red teaming, and RL environments.

75Turing

Expert data and verifiable tasks for coding and computer-use agents.

76Ulam

Mathematical reasoning data, private evaluations, and RL gyms.

77Veris AI

Enterprise simulations for testing and training AI agents.

78Verita AI

Expert sourcing and operations for annotation and RL projects.

79Vetto AI

Expert data, RL environments, and evaluations across modalities.

80Vmax

RL tooling that turns proprietary data and evaluations into environments.