Agent evals night

​Agents break in ways unit tests don't catch.

​They can pass the demo and still pick the wrong tool, misread an MCP server's schema, or lose track of what it learned three steps earlier. Evals are how teams catch these failures before they reach production. The tooling is still young, so plenty of teams build their own harnesses, LLM judges, and trace reviews from scratch.

​Join us for an evening of lightning talks and live demos from teams who run evals on agents every day. We'll cover why agents pick one product over another, how to test one MCP server across the different clients people connect it to, and how to measure what an agent learns in context.

​Speakers

• ​Robert-George Crupa, AI Engineer at Apify, on how we evaluates 100k+ tools

• ​Joshua Chung and Vivek Shukla, co-founders of Xulux, on evals for in-context learning

• ​Renzo Viale, CEO of Oqoqo, on why agents choose (or ignore) your product

• ​Prathmesh Patel, CEO of MCPJam, on cross-client evals for MCP servers

• ​More speakers announced soon

​What to expect

• ​10 to 15 minute lightning talks and live demos on agent evals

• ​Food and drinks

• ​Networking with AI engineers and agent builders

​Agenda

• ​6:00 pm: Doors open, food and drinks

• ​6:30 pm: Lightning talks and live demos

• ​7:45 pm: Networking

• ​9:00 pm: Wrap

​Good to know

• ​Free to attend. Spots are limited, so every RSVP needs host approval.

• ​Apify will take photos and short clips at the venue for social media.

​About Apify

​Apify is the largest marketplace of tools for AI: thousands of ready-to-run tools that agents and builders use to get more value from the web.