In detail
- Founded 2023 by Anand Kannappan and Rebecca Qian; total funding now $70 million.
- Uses 'digital world models'—simulations of websites and internal systems—to train and evaluate agents via reinforcement learning.
- Revenue grew 15-fold over the past year; customers include virtually all frontier AI labs and many startups.
- Series B led by Greenfield Partners, with participation from Notable Capital, Lightspeed, Datadog, and Samsung.
Why it matters
This addresses a critical gap: benchmarks alone don't prove agents can reliably execute complex, real-world tasks. For enterprises deploying AI agents, such validation is becoming a business-critical risk.
For you If you're deploying AI agents for mission-critical processes (bookings, financial analysis), understand how your provider tests them—Patronus' approach is becoming the standard.