Find companies working with business data

Abundant builds reinforcement-learning environments, evaluation benchmarks, and training datasets for AI labs, AI startups, and large enterprises developing and improving AI agents.

AfterQuery sells expert-created and synthetic training datasets, reinforcement-learning environments, and custom evaluations to AI labs and enterprises developing foundation models and agents.

Alice, formerly ActiveFence, sells AI safety and security training datasets, agentic RL environments, and expert-led red teaming and evaluations to foundation model labs; it also sells WonderSuite guardrail software to enterprises and content moderation to online platforms.

Anthromind supplies AI developers, research teams, and businesses with expert-curated training and preference data, custom LLM evaluations, and fine-tuning support through tailored projects and managed data-curation services.

Apex Data Sciences, the AI division Apex CoVantage launched in 2024, sells expert RLHF and SFT data, expert-built benchmarks and image, text, video and lidar annotation to AI teams. Its parent also sells publishing, accessibility and healthcare billing services.

Appen supplies AI labs and product teams with human-produced training data, annotation, expert feedback, RL environments, and model evaluations for machine-learning and agentic AI systems.

Arena sells AI model evaluations built on real-world human feedback to labs, enterprises and developers, with analytics, feedback samples and SLA-backed delivery. Its privacy policy says it may share users' prompts, votes and other content with AI model developers to evaluate, improve and develop their products; its FAQ says such shared data is de-identified.

Builds production-style simulated workspaces and verifiable benchmarks for AI agents, helping AI developers train and test long-horizon tasks in cybersecurity, incident response, software engineering, and tool use.

Aviro sells reinforcement-learning environments, post-training datasets, and evaluations for AI labs and teams building long-running agents, with tasks spanning coding, computer use, and enterprise knowledge work.

Claru provides custom and licensed video, robotics demonstrations, and human annotations to AI labs building robotics, embodied AI, and world models.

Abu Dhabi company that sources, labels and evaluates training data for AI labs, hyperscalers and robotics teams, including human-ranked model responses, expert-written evaluation sets, Arabic speech data and first-person video of real work filmed by paid contributors through its KLKT app, and also sells its own Arabic speech models and voice agents.

CueBench builds reinforcement-learning environments and task benchmarks for AI labs and developers to train and assess frontier models on difficult scientific-reasoning and performance-engineering problems.

Datacurve sells expert-built coding data, reinforcement-learning environments, benchmarks, and evaluations to foundation-model labs and AI developer-tool companies.

DataOcean AI sells off-the-shelf datasets and custom data collection, labeling, and model evaluation services to AI developers and enterprises, supporting LLM, speech, vision, multimodal, and autonomous-driving systems.

Deccan AI sells AI labs and developers expert-created post-training datasets, human feedback, reinforcement-learning environments and evaluations; it also offers enterprises monitoring and AI-enabled operations tools.

Diffuse Labs builds reinforcement-learning environments for frontier AI labs, generating scalable tasks and verifiable rewards so developers can train agents and measure gains across coding, chip design, Kubernetes operations, and enterprise software.

d_model builds RL environments where agents learn to do interpretability and alignment research, and runs similar agents in its own research.

Emulated offers AI teams post-training datasets, production-like software-engineering environments, and benchmarks to train and evaluate agents on coding, infrastructure operations, machine-learning engineering, and research tasks.

Supplies large-scale pretraining and post-training datasets, including code and web, document, and video data, plus verifiable STEM RL environments and custom data sourcing for frontier AI labs.

FoundryAI creates expert-reviewed coding and other technical-workflow data, RL-ready task suites, private model evaluations, and regression tests for AI development teams.

G2i supplies coding post-training data to AI labs (RL environments, evals, RLHF preference data, SFT datasets, scrubbed seed data) through managed projects or contract engineers placed on lab teams. It also runs an engineer hiring platform and tech conferences.

Hiloop sells automated research infrastructure to AI/ML teams, running experiments across their models, data, training methods, prompts, tools, and systems against customer-defined evaluations, then returning verified improvements.

Huzzle Labs sells reinforcement-learning environments, expert-produced training data, evaluations, and custom-trained models to AI labs and enterprises seeking agents that can handle real software and business workflows.

Idler sells reinforcement-learning environments, benchmarks and evaluation datasets based on production work to AI labs and other teams training and evaluating models.

Innodata provides AI developers and enterprises with training data, human feedback, annotation, reinforcement-learning environments, model evaluations, and red-teaming services for building and improving AI models and agents.

Intelligence runs DesignArena, where people vote between AI-generated websites, games, images, video and audio, and sells model evaluations to AI labs. It says it shares users' prompts, outputs and votes with AI model developers, who may train on them, after taking steps to remove identifiers like usernames; its terms let it charge for that use.

Invisible sells expert-created training data, model fine-tuning, evaluations, and custom reinforcement-learning environments to AI labs and developers; it also provides AI software and implementation support for enterprises.

Labelbox sells simulated reinforcement-learning environments, expert-generated preference and training data, and evaluation services to frontier AI labs; it also offers enterprise teams software for running agents on real company workflows.

Markov sells computer-use datasets and task environments—including expert CAD, productivity, coding, and gaming workflows—to AI labs and developers training or evaluating desktop agents.

Mechanize builds simulated software-engineering tasks and benchmarks for frontier AI labs, helping them train and evaluate coding agents using graded performance.

Mercor sells expert-created training data, reinforcement-learning environments, and evaluations and benchmarks to frontier AI labs and other AI developers. It also offers enterprise data licensing and evaluation services.

Metaphi sells custom and reusable simulated work environments, training datasets, and agent evaluations to frontier AI labs and enterprise AI teams, helping them train and assess agents on realistic software, finance, and content-generation tasks.

micro1 provides expert human data and post-training services, real-world RL environments, robotics data, and AI-agent evaluations for frontier AI labs, enterprises, and robotics teams.

Miro Advisory helps established businesses license their operating history (support tickets, reconciliations, dispatches, approvals and other workflow records) and software companies license private codebases with their commit, issue and review history, to AI developers for training and evaluation. It acts as an advisor and intermediary: it handles qualification, rights review, buyer discussions and the licensing process.

Navaday supplies expert-created training data and model evaluations across professional and technical fields. It manages specialists who write examples, correct existing data, compare model answers and produce held-out test sets. It also prepares and helps license business records for AI development.

Olam Labs builds simulated multi-agent environments, behavioral evaluations and game-derived datasets for AI labs and researchers training or evaluating models’ social, agentic and safety capabilities.

Pareto supplies AI labs with specialist-created training data, evaluations, and RL environments that turn expert judgment into signals for assessing and improving frontier models.

Patronus AI builds training data and simulated environments to train and improve AI models and agents. Its research studies model failures in alignment, coding, knowledge work, science and computer use, and it publishes benchmarks.

Petrarch sources operational records from companies under signed agreements, de-identifies them with its in-house Lindo software inside the supplier's environment, and licenses or sells the prepared datasets to AI labs and applied AI startups as training data.

Phinity builds chip-design reinforcement-learning environments and training infrastructure for AI labs, while developing agents to design, verify, and eventually tape out custom chips.

Builds simulated environments, tasks, and verifiers for AI labs to train and evaluate long-horizon agents; its Horizon-SWE benchmark tests end-to-end software engineering workflows.

pre.dev sells runnable, graded coding environments and training and evaluation data built from licensed private software repositories to AI labs and developers training coding agents.

Preference Model builds reinforcement-learning environments for frontier AI labs, helping train models to handle machine-learning research and engineering tasks.

Prime Intellect sells AI labs and developers an open stack for reinforcement-learning environments, model evaluations, hosted training, inference, sandboxes, and GPU compute to build and improve agentic models.

Proximal builds coding tasks, training data and RL environments that AI labs and enterprises use to train software-engineering agents.

Quesma publishes coding-agent benchmarks and builds RL training environments for frontier labs; it also markets tools that help development teams understand agent costs, workflows and outcomes.

ReasonCore provides expert-curated AI training datasets, reinforcement-learning environments, and adaptive evaluations for frontier AI labs, emerging AI labs, and enterprises developing or deploying advanced models.

Refresh sells simulated software environments and verifiable evaluations for training and testing AI agents, alongside real-workflow coding datasets, to frontier AI labs and enterprises.

Replay pays companies for their work history (code, internal docs, support tickets, CRM and financial records), strips personal identifiers and licenses the cleaned datasets to frontier AI and data labs. The same company runs Sunset, a startup wind-down service.

Revelo sells AI labs code-focused training data, evaluations, benchmarks, and RL environments, with expert-led data creation and managed workflows to help improve coding models and agents.

RL Labs builds long-horizon software-engineering environments where frontier AI labs can train models on complex, multi-step coding tasks.

Runloop provides isolated development sandboxes, orchestration, and benchmarking tools so AI developers and labs can build, run, score, and refine coding agents, including reinforcement-learning and fine-tuning workflows.

San Francisco Data prepares and licenses training data derived from companies’ production AI traces, starting with coding agents. Its research work turns recorded interactions into reproducible tasks and training environments for frontier models.

Scale sells AI labs and businesses custom training data, human feedback, annotation, RL environments, and model evaluations, alongside software to build, test, and deploy AI agents and applications.

San Francisco company that builds finance and coding RL environments, with tasks created by private equity professionals and senior software engineers, and sells ready-made environments, custom tasks, model evaluations and human preference data to frontier AI labs.

SimpleClosure’s AssetHub connects startups with AI buyers interested in licensing or acquiring source code and workspace data.

Snorkel sells expert-built training datasets, agent environments, benchmarks and evaluations—through ready-to-use products and custom data-development services—to frontier AI labs and enterprise AI teams.

SuperAnnotate sells AI data software and managed expert services to AI labs and developers, covering multimodal annotation, training data, model and agent evaluations, and reinforcement-learning environments.

Surge sells AI labs and developers expert-built training data, human feedback, evaluation benchmarks, and interactive reinforcement-learning environments to train and assess AI models and agents.

Symbal supplies AI labs with multimodal post-training datasets, private evaluations and benchmarks, and human-in-the-loop data; it also builds enterprise agents using customers’ first-party workflows.

Telegraph Lab sources and licenses business records for AI training. It reviews data rights, prepares datasets through de-identification and structuring, and arranges delivery to AI labs. Its sources include workplace communications, business-system records, company files and operating histories.

TELUS Digital supplies AI labs and developers with expert-built training data, annotations, agentic RL environments, red-team testing and evaluations, plus custom multimodal and physical-AI datasets and managed data-collection services.

Theta sells custom simulated work environments and hands-on implementation to AI labs and enterprises, helping them train and evaluate specialized agents on realistic, multi-step business workflows.

Toloka supplies AI labs and developers with human-expert and synthetic training data, agent environments, evaluations, and red-teaming to build and assess AI models and agents.

Turing supplies AI labs with training data, evaluations, and RL environments for coding, enterprise knowledge work, and STEM. It also helps large enterprises build and deploy AI agents in important workflows.

Uber AI Solutions provides custom AI data collection and annotation, specialist human feedback and RLHF, red-teaming, and model evaluation for frontier labs, enterprises, robotics firms, and autonomous-vehicle developers.

Ulam sells reasoning datasets, expert-reviewed traces, benchmarks, and verifier-backed RL environments—primarily for mathematics and STEM—to AI labs and developers.

Vals AI sells evaluation software, expert-built benchmarks, and review tools to AI labs and enterprise engineering teams testing models and AI applications on practical, domain-specific tasks.

Veris sells AI developers a simulation and reinforcement-learning platform that recreates enterprise tools, systems, and users, helping teams train, test, and improve agents on realistic workflows before production.

Vetto sells AI labs and developers expert post-training data, human evaluation, custom RL environments, and verifier-graded benchmarks to help train and assess models on practical tasks.

Vmax markets Campaign, a self-play task-generation and RL platform that uses developers’ proprietary data and evaluations to create training environments and refine agents on new task examples.