Agent Harness Engineering: Everything Around the Model.

About the track

Most agent failures in production aren’t model failures. They’re harness failures: the tools, permissions, loops, and guardrails around the model that nobody treats as a discipline yet. We shipped agents into production faster than we built the engineering to keep them reliable.

Engineers running agents in production past the prototype, under real constraints like regulated data, legacy systems, and hard cost ceilings.

What we aim to cover

  • Working, episodic, semantic, and procedural memory
  • Memory-writing, retrieval, consolidation, and forgetting policies
  • Context management across sessions, external stores, and learned/adaptive memory
  • Evaluating memory: recall, relevance, consistency, and task performance
  • Stale memories, conflicting information, privacy, and permissions

Every talk comes from something the speaker built and measured, including the failure modes that forced a redesign. War stories and trade-offs.

Attendees will walk away with concrete harness patterns you can put into your own agents the same week.

Track host

Tony Kipkemboi

Staff Engineer, AI Operations Guild. Prev, Founding DevRel Engineer & Partnerships Lead , CrewAI

Tony Kipkemboi is an AI engineer and educator focused on building enterprise AI agents and teaching developers how to ship them. He’s currently building Rekody and publishing widely used technical guides on agentic systems, fine-tuning, and AI infrastructure. Previously, he led developer advocacy for AI agents at CrewAI and held roles at Snowflake (Streamlit) and Bloomberg, giving him a front-row view of how agent frameworks and data tooling are adopted inside large organizations.

His talks have reached developers through PyCon US, ODSC, IBM TechXchange, O’Reilly live events, and the Harvard Kennedy School. His work centers on making enterprise agent deployment practical: what to build, how to evaluate it, and how to get it into production.

Summit (2 days).

MLOps World | GenAI Summit 2026 is a two days of case studies, workshops, and expo on taking AI/ML and agentic systems into production – at the Etter-Harbin Alumni photo – full-bleed hero or browse files.

Share