Patronus AI, a San Francisco-based company building simulation infrastructure for AI agents, has raised $50 million in Series B funding led by Greenfield Partners. The company develops Digital World Models, language diffusion models that simulate digital environments to train and stress-test autonomous AI agents before deployment. The capital will fund compute and infrastructure to scale these models.
Simulation Infrastructure Gains Momentum
The timing comes amid a wave of investment in AI reliability tools. Arize AI raised $70M Series C in February 2025, while Fiddler AI secured $30M Series C in January 2026. Patronus AI's approach focuses on pre-deployment simulation rather than post-deployment monitoring.
Static Benchmarks Fall Short for Agents
AI agents fail 63% of the time on complex tasks, according to recent analysis. Current static benchmarks cannot capture real-world behaviors like recovering from errors or handling ambiguous instructions across long workflows. Enterprises deploying agents in finance, healthcare, and software development lack ways to verify reliability before production use.
Digital World Models Enable Dynamic Training
Patronus AI builds language diffusion world models that predict environment behaviors and steer agent actions in digital workflows such as coding, GUI navigation, and tool use. Its Percival tool evaluates multi-step agent workflows and detects over 20 failure modes. This differs from competitors focused on observability after models reach production.
As Anand Kannappan, CEO, noted:
"Benchmarks only provide static evaluations that show if a model can perform in a tightly controlled setting. They do not tell you whether an agent can navigate ambiguity, recover from failure or operate reliably across long, unpredictable workflows."
Greenfield Partners Leads Strategic Round
Greenfield Partners led the round with participation from Lightspeed Venture Partners, Notable Capital, Datadog, and Samsung Ventures. The syndicate reflects conviction in simulation as foundational infrastructure, with strategic investors from observability and hardware adding ecosystem validation. Total funding now reaches $70 million.
AI Evaluation Market Expands Rapidly
The AI Model Evaluation Platform market stands at $2.8 billion in 2025 and the LLM security market at $2.27 billion, both growing above 20% CAGR. per market research and security analysis. Patronus AI reported 15x revenue growth in the past year while serving frontier labs and enterprises including Etsy and CARIAD.
Aggressive Scaling Underway
The company plans to invest the new capital in compute and infrastructure for training Digital World Models at scale. It recently opened a San Francisco headquarters and launched a new graduate hiring program as its research team expands rapidly.