Patronus AI
About
Patronus AI builds simulation infrastructure for training and evaluating AI agents. Its work includes Digital World Models and generative RL environments for realistic software and tool workflows, alongside long-horizon benchmarks, reward systems, training datasets, and agentic supervision tools such as Percival.
What They Offer
Commercial or operational services this company provides.
Products & Public Artifacts
Open-source evaluation model, released with evaluation code and test sets.
Visit artifact →Digital world model for dynamically generating realistic software and tool environments for agent training.
Visit artifact →Simulation framework for generating dynamic environments and interactions for agent training and evaluation.
Visit artifact →Human-workflow training dataset built from realistic Figma design activity.
Visit artifact →Long-horizon benchmark for measuring frontier-agent performance on complex tasks.
Visit artifact →Benchmark for agentic evaluation across tool-using workflows.
Visit artifact →Agentic supervision and evaluation tooling for monitoring and scoring AI systems.
Visit artifact →Technical Capabilities
Only publicly documented capabilities are shown. Absence does not imply the capability is unavailable.