Reinforcement Learning Startups

rct AI develops artificial intelligence systems for non-player characters in video games and metaverse environments, using deep learning and reinforcement learning to create interactive and adaptive gameplay experiences.

5 investors

Sepal AI generates training and evaluation datasets for AI models at scale, working with leading AI research labs and enterprises to improve model performance on real-world tasks.

5 investors

Osmosis provides reinforcement learning infrastructure that enables AI agents to improve themselves from experience. The platform allows developers to fine-tune models on proprietary data with full ownership, addressing practical challenges in deploying and scaling AI agents across multi-step tasks.

4 investors

Theta builds simulation environments and training infrastructure for AI agents specialized to perform enterprise workplace tasks. The company partners with enterprises and research labs to create custom agents that master domain-specific workflows, tools, and decision-making across domains like finance, back office operations, and software development.

3 investors

TrainLoop is a post-training research lab that trains specialized AI models for enterprise customers in pharma, biotech, logistics, and banking. The company develops custom foundation models and agents optimized for long-horizon tasks and mission-critical workflows in high-risk industries.

3 investors

Agentin builds an AI system that automates revenue and spend workflows across enterprise software systems like SAP, Salesforce, and NetSuite, replacing manual coordination with autonomous agents that monitor and fix financial transactions.

2 investors

Atmeto develops machine learning technology for clean energy generation, aiming to decouple carbon dioxide emissions from energy production.

2 investors

Aviro builds training environments and simulated spaces for long-horizon agents and frontier AI models to learn complex navigation and tool use.

2 investors

Cortex Robot builds datasets for embodied AI and robotics, including egocentric video data from real workplaces and robot trajectories for training physical intelligence systems. The company serves robotics developers and AI researchers building foundation models and world models for embodied tasks.

2 investors

HUD builds evaluation tools and reinforcement learning environments for computer-use AI agents, helping frontier AI labs test and train agents across diverse tasks at scale.

2 investors

A company focused on growth and productivity solutions that prioritizes customer needs over technology for its own sake.

2 investors

Traverse produces training data for frontier AI models to develop judgment and reasoning in ambiguous, non-deterministic tasks like law, healthcare, and strategic decision-making. The company partners with AI labs to capture real expert reasoning patterns that enable models to perform complex white-collar work at superhuman levels.

2 investors

Vibrant Labs builds tools for teams to develop AI systems that function effectively in real-world applications. The company focuses on providing developer tooling for machine learning and AI implementation.

2 investors

Cartpole develops simulation environments designed to train advanced machine learning models using reinforcement learning techniques.

1 investor

Coasty provides AI agents that automate healthcare revenue cycle tasks like eligibility verification, prior authorization, claim submission, and denial management by performing repetitive work across EHRs, practice management systems, and payer portals through screen interaction. The platform is designed for healthcare billing teams and operates within customer environments with full audit trails and approval gates for payer-facing submissions.

1 investor

EdotEnv is a quant research lab building reinforcement-learning environments at scale from financial market data, designed to increase in difficulty so AI agents can continuously improve. It works with frontier AI labs and academic groups on research harnesses, evaluation benchmarks, and post-training environments toward recursive self-improvement.

1 investor

Enact provides post-training data for robotics and physical AI models, running policies on tasks to find failure states and generating targeted demonstration and recovery datasets. It also evaluates model performance in-house to confirm the data resolves those failures.

1 investor

Fabraix develops automated security testing software that probes AI agents for vulnerabilities using adaptive attack techniques, enabling companies to continuously validate the safety of customer-facing AI systems without manual penetration testing.

1 investor

Flash is a platform that trains custom machine learning models for AI agents using user-provided data. It offers fixed-price training runs with downloadable model weights that can be deployed anywhere, positioned as a cheaper and faster alternative to frontier models for routine inference tasks.

1 investor

Hillclimb develops training data for recursive self-improvement in AI models, focusing on aggregating human research data and automating reinforcement learning environment creation. The company sells the resulting data to frontier AI laboratories.

1 investor

hiloop provides snapshottable, forkable compute infrastructure with built-in observability for AI agents that propose, run, and verify experiments (autoresearch). It offers self-provisioning sandboxes via API, brokered secrets, and per-run tracing with token and cost tracking queryable via SQL, aimed at teams doing autoresearch, model training, and performance-critical development.

1 investor

Idler develops training environments for reinforcement learning that enable AI models to learn programming through simulation of real-world coding tasks.

1 investor

Kairos provides an applied AI layer that enables enterprises to build specialized agents trained on their proprietary data and workflows. The platform helps critical industries bridge the reliability gap in enterprise AI by embedding operator expertise into customized agents that work within messy real-world business environments.

1 investor

Magma helps companies that build AI agents package and license their agents' interaction data — traces, corrections, and completed workflows — to AI labs, paying those companies a revenue share. It handles onboarding, data mapping, and packaging so agent logs that would otherwise be discarded can be monetized.

1 investor