Honeycomb Observability Day
6–7 October 2026
Convene Bishopsgate, London
Real talk from real practitioners
The rules of production are changing. AI is introducing complexity that traditional observability was never designed for, and the teams navigating it best are the ones who understand their systems most deeply.
O11yDay London is back, bigger than ever.
Join developers and engineering leaders sharing what's actually working in the agent era — real challenges, hard-won lessons, honest outcomes.
6th October — Hands-on workshops
7th October — A full day of practitioner-led talks
A room full of people who get it.
Sam Newman —
Cobots, dark factories and architecting for AI
Keynote
Author of Building Microservices and Monolith to Microservices, Sam has spent years helping organisations understand how their systems actually behave in production.
His keynote explores the hidden complexity accumulating inside AI-augmented systems and what we need to do about it.
Corey Quinn —
Nobody Knows What Anything Costs Anymore
Corey is Chief Cloud Economist at The Duckbill Group and one of the most sharp-tongued voices in cloud infrastructure.
He’ll be unpacking the real cost structures emerging as AI workloads hit production, because tokens aren’t free, and pretending otherwise is expensive.
Yana Chen — You can't vibe code uptime... or can you?
Yana is a Software Engineer at Baseten (a cloud scale inference platform with customers like Notion, Cursor and Lovable). For them, inference reliability isn’t a nice-to-have, it’s the product.
She'll be sharing hard-won lessons from running ML inference at scale: what breaks, what doesn't, and how you actually debug it.
Brian Scanlan & Kesha Mykhailov —
Operational Excellence for Agents with fin.ai
Your agent passed its evals, but that doesn't mean it's working. Also, you're now an AI engineer whether you like it or not.
At Fin (formerly Intercom), hundreds of engineers and even more outside of engineering use Claude Code every day to accelerate their work. Brian will show how they instrument Claude Code with OpenTelemetry and session transcripts, and what they do with the data: personal and team developer experience dashboards, experiments on cost and effectiveness, hunting down cost and effectiveness outliers, and working out why one person's setup is screwy.
Our AI code reviewer, inexplicly called Shrek (Kesha will explain in detail why this name is important), reviews thousands of pull requests a week and approves plenty of them on its own. Kesha will tell the story of a model change that passed our evals but blew up in production.
Event details
Venue
Convene, 22 Bishopsgate, London
Day 1 - Tuesday 6 October
Workshops: Hands-on, group sessions for engineers who want to go deep.
Day 2 — Wednesday 7 October
Conference: A single track of practitioner talks. No fluff.