Yash Thakare

Hackathon to Reality: What Breaks When You Build AI Agents Fast (and How to Fix It)

AI agents are easy to demo,but hard to ship. In this session, I’ll share real lessons learned from building AI agents under hackathon-style constraints and pushing them toward real-world use. We’ll look beyond the “happy path” to explore what actually breaks when agents move from prototype to practice. The talk covers common failure modes such as hallucinated tool usage, unreliable orchestration, cost blowups, prompt drift, and poor observability — and the practical techniques used to mitigate them. I’ll walk through concrete architecture decisions, guardrails, and evaluation strategies that helped stabilize agent behavior. The session includes demos, failure examples, and reusable patterns that developers can apply when building or integrating AI agents into real workflows. Attendees will leave with a clearer understanding of how to design AI agents that are not just impressive in demos, but resilient, debuggable, and production-ready.