What is Janus (YC X25)?
Janus battle-tests your AI agents to surface hallucinations, rule violations, and tool-call/performance failures. We run thousands of AI simulations against your chat/voice agents and offer custom evals for further model improvement.
Problem
Users are manually testing AI agents for hallucinations, rule violations, and performance failures, which is time-consuming, inconsistent, and fails to scale.
Solution
An automated simulation testing platform where users can deploy AI agents to undergo thousands of simulated interactions, identify vulnerabilities, and refine models with custom evaluations (e.g., testing chatbots against edge cases).
Customers
AI developers, machine learning engineers, and teams building chat/voice agents (e.g., customer support bots, AI companions).
Unique Features
Scalable adversarial testing framework, custom evaluation metrics, and detailed failure analysis reports for model iteration.
User Comments
Saves weeks of manual testing
Uncovered critical edge cases we missed
Essential for production-ready AI
Custom eval system is powerful
YC-backed reliability
Traction
Launched on Product Hunt (May 2024), part of Y Combinator W24 batch. Traction details like MRR/users not publicly disclosed.
Market Size
The AI quality management market is projected to reach $1.3 billion by 2027 (MarketsandMarkets, 2023).
