Harness Engineering: From Capable Models to Reliable Agents
A capable model alone doesn’t make a reliable agent. Production systems need a harness that manages state, supplies relevant instructions and skills, governs tool use, and handles failures.
In this session, Nicole Koenigstein, author of Harness Engineering (O’Reilly), introduces the discipline through a practical example, connecting state management, reusable skills, and verification and repair loops. We’ll explore how these components shape agent behavior within defined operational boundaries, and how execution traces and evaluation guide improvements.
You’ll leave with a clearer understanding of the harness’s responsibilities and how to apply these principles when building agents for production.
Ali Parandeh, author of Building Reliable AI Agents Series (Manning) will chair and moderate the Q&A.
This session is for engineers who build or deploy AI agents, or are about to, and want a clear picture of what the harness is responsible for.
Date: Tuesday 27 October 2026
Time: 6pm GMT / 7pm CET
Location: Online. Joining link sent after registration.
We’ll cover:
What a harness is responsible for, and where it ends and the model begins
How state management and reusable skills affect each other
Verification and repair loops, and what they catch that prompts do not
Using execution traces and evaluation to decide what to change
Setting operational boundaries for tool use
No coding setup required.
✨ Special for participants: Participants of the webinar will receive the following resources:
A Harness Engineering Cheat Sheet as a free resource to accompany the webinar, showing how the different parts of the harness affect each other and why they matter.
Agent Requirements Document (ARD) Template: Use this to plan your agents for functional, performance, reliability and security requirements. Give it to Claude Code or Codex to help you think through the design of your bespoke agents and the required harnesses.
Speakers:
Nicole Koenigstein is an AI researcher and practitioner in agentic systems, working across research, consulting, teaching and system implementation. She is the author of AI Agents: The Definitive Guide, Transformers: The Definitive Guide and Harness Engineering (O’Reilly), and Math for Machine Learning and Transformers in Action (Manning). She has served as an external evaluator for a European Commission AI Grand Challenge and advised IOSCO on generative AI in regulated environments.
Ali Parandeh is a Chartered Engineer and AI software engineer. He is the author of Building Generative AI Services with FastAPI (O’Reilly, 2025) and is writing the Building Reliable AI Agents series for Manning. Through Build Your AI he helps engineering organisations in rail, aerospace and infrastructure take AI from pilot to production. His background includes engineering work on projects for the European Space Agency, HS2 and Network Rail.
Agenda:
6pm to 6:05pm: Introductions and context (Ali)
6:05pm to 6:50: Talk (Nicole)
6:50pm to 7pm Q&A