Harnesses in AI: A Deep Dive
About this event
Abstract
The agent hit a login page, panicked, reported success anyway, and the upvote never happened. Tejas Kumar's diagnosis: not a prompt problem. A harness problem.
The demo builds a browser agent on GPT-3.5 Turbo (consciously choosing a VERY old model to show how good harness eng can improve it a lot) against Hacker News and layers in a harness without touching the prompt once. Guardrails cap iterations and compact context. A verify step reads the tool call history to catch the agent lying about what it did. A login handler watches the browser URL each loop and injects credentials programmatically when it hits the login page. By the end the cheap old model reliably logs in and upvotes the post.
About the Speaker
Tejas Kumar is an international keynote speaker with an engineering background spanning 22 years, from design to front-end to back-end to DevOps. He has been writing code since age 8 and worked either full-time or contractually for Spotify, Vercel, G2i, and others, picking up things along the way.
Today, Tejas shares talks at large with developer communities worldwide, equipping them to do their best work.
Source: meetup