From simulation to support: Using evals to launch dependable AI agents

​AI agents are moving from demos to production, but reliability remains the hardest challenge. Without rigorous testing, they risk regressions, unpredictable behavior, and broken customer experiences.

Most teams can get an AI agent to demo well. The hard part is making sure it performs consistently in production. In this workshop, Brooke Hopkins, co-founder & CEO of Coval (YC W25), will share how simulation and evaluation can help teams go from one-off successes to reliable, scalable AI support agents.

You’ll learn:

• Why simulation and evaluation are critical for scaling AI agents beyond the demo stage
• How startups and enterprises use evals to prevent regressions and ensure consistent support quality
• The future of testing frameworks as AI agents become more autonomous and multimodal

We’ll close with a live Q&A with Brooke, diving into the lessons from Coval’s journey and the path to building agents customers can truly trust.
Perfect for: Founders, AI ops teams, engineers and support leaders who want to move past one-off successes and confidently scale AI-powered support.

Featured speakers: Brooke Hopkins (CEO & co-founder @ Coval), Marie Schneegans (CEO & co-founder @ 14.ai), and Maya Czeneszew (Founding Engineer @ 14.ai)