Keynote Speakers

Amara Okonkwo-Reyes
Chief AI Officer
Lanternworks
Amara has put AI in front of two million customers and still reads the failure reports herself.

Dr Tomasz Wieliczko
Director of Research
Halden Institute for Machine Reasoning
Tomasz has studied automation for thirty years and thinks most of the new problems are old ones.
13 to 14 May 2027. Two days on putting AI to work.
MAY 2027
13 - 14

Amara Okonkwo-Reyes
Chief AI Officer

Dr Tomasz Wieliczko
Director of Research

Meera Vaidyanathan
Head of AI Design

Callum Ferreira-Hughes
Staff Engineer, Agents

Yuki Tanabe
Founder

Nadia El-Masri
VP of Machine Learning Engineering

Samuel Adeyemi-Clarke
Director of Evaluation

Ingrid Solheim
Principal Researcher

Mateo Villanueva
Co-founder

Leilani Kahale
Head of AI Operations

Rohan Mehta-Lindqvist
Author

Eleanor Whitcombe
Host and Editor
What two days gets you
One stage, so nobody misses anything.
One stage, no tracks
Every talk is for everyone. No picking between rooms, no fear of missing the good one.
Practice, not predictions
Every speaker runs AI in production. Expect numbers, failures and what they changed.
A workshop seat
Leave with the rules for an AI system you look after written on one page.
The evening on the terrace
Speakers stay. Food, drinks and no agenda until the light goes.
Two days, in order
Day one is building it. Day two is running it.
Day 1
- 09:3009:50
Opening: the year in AI, without the hype
Eleanor WhitcombeYour host sets out the two days: one on building AI people can rely on, one on running it.
Main stage - 09:5010:35
From demo to dependable
Amara Okonkwo-ReyesWhat changes between a demo that impresses and a system two million customers lean on, and the decisions that only pay off in year two.
Main stagekeynote - 10:4511:20
Designing for a system that can be wrong
Meera VaidyanathanHow to show uncertainty without frightening people, and every place Quietloop decided to say "I am not sure".
Main stage - 11:4012:15
Small models, sharp tools
Yuki TanabeWhy a four person studio ships narrow models that run on one machine, and an honest count of what that has cost.
Main stage - 12:1512:55
What teams still use in month three
Ingrid SolheimFindings from going back to the same teams months after an AI tool arrived. What stuck, what quietly disappeared.
Main stage - 12:5514:00
Lunch
A long lunch, on purpose. Seasonal food, no laptops on the tables.
Garden room - 14:0014:35
Evaluation on one page
Samuel Adeyemi-ClarkeThe single printed page of tests that replaced a forty page evaluation suite, and the arguments it ended.
Main stage - 14:4015:15
The data nobody wanted to label
Mateo VillanuevaSix years of collecting and checking data before training anything. What it takes, who pays for it, and how to tell early from wrong.
Main stage - 15:4016:25
Fireside: the patient machine
Rohan Mehta-LindqvistRohan reads three short chapters and Eleanor argues with one of them. Questions from the floor.
Main stage - 17:0019:00
Evening on the terrace
Drinks, food and no agenda. Speakers stay for the evening.
Roof terrace
Day 2
- 09:3010:15
Live build: an agent that ships a pull request
Callum Ferreira-HughesA real coding agent, given a real ticket on stage, from first prompt to a reviewed and merged change with the tests green.
Main stage - 10:2010:55
Boring model releases
Nadia El-MasriFrom one model release a quarter to many a week, and the unglamorous work of making that dull.
Main stage - 11:2011:55
On call for agents
Leilani KahaleRunning agents that move real money: the rota, the runbooks and the stop button every agent must have.
Main stage - 12:0012:45
Panel: when to let the agent decide
Eleanor WhitcombeAmara Okonkwo-Reyes, Nadia El-Masri and Yuki Tanabe on which decisions to hand over and which to keep. Hosted by Eleanor Whitcombe.
Main stage - 12:4513:50
Lunch
Same long lunch, different menu.
Garden room - 13:5014:50
Workshop: write your agent's rules on one page
Samuel Adeyemi-ClarkeBring an AI system you look after. Leave with what it may and may not do written on one page. With Meera Vaidyanathan. Places are limited.
Workshop room - 15:1516:00
What thirty years of automation taught us
Dr Tomasz WieliczkoThree automated systems older than most of the audience, taken apart to show where the people were kept, and why it mattered.
Main stagekeynote - 16:0016:20
Closing: what to do on Monday
Eleanor WhitcombeOne thing to automate and one thing to keep in human hands. Then the doors.
Main stage
Companies putting AI to work.
Made possible by
In Partnership With
















Tickets
Online Pass
Watch the main stage live from wherever you are.
- Main stage live stream, both days
- Live questions to the host
- The summit briefing by email
Summit Pass
Both days in the room, with lunch, a workshop seat and the evening on the terrace.
- Both days, every session
- Lunch and coffee on both days
- The evening on the terrace
- A seat in one workshop
- Recordings afterwards
Not ready to book?
Annual AI Summit is a two day conference for the people who build, deploy and run AI in real organisations. Twelve speakers, one stage, and an honest look at what works once the demo is over.
Good questions
Short answers.









