AI Agents: What the Builders are Learning
Event by Dtechtive
Wed, Oct 28, 2026, 2:00 PM – 2:45 PM

INTRODUCTION

Many organisations are now deciding whether to let AI agents, AI that takes actions as well as answering questions, work inside their systems. The people building those agents, and the models and tools underneath them, are often the first to see what works and what doesn’t. What they learn rarely makes it into the headlines.

So what does AI-readiness look like from the builder’s side? This is the third episode in a series exploring what AI-readiness means across industries, each time from a different vantage point. In Episode 3, Amanda Brock shares what she is hearing from the developers, start-ups and open source communities building AI: what they are finding hard, what open should mean for an AI model, who is responsible when an agent gets something wrong, and why well-described data and shared standards make agents more reliable.

SCHEDULE

Theme: AI-Readiness | Episode 3

The audience may add comments or questions during the session and we will take as many as time allows. The session stays available afterwards if you cannot join live.

GUEST

Amanda Brock. CEO, OpenUK.

Amanda leads OpenUK, the not-for-profit organisation for open technology in the UK, covering open source software, hardware, data, standards and AI. Since 2019 she has built it into a think tank and advocacy organisation on open source and AI that works with businesses, developers, governments and international bodies. In 2025 she founded OpenHQ to help open source communities take part in policy and governance, and OpenUK’s State of Open Con brings builders and policymakers together, this year on the road in Edinburgh and Belfast. She also advises technology start-ups and scale-ups, including in cybersecurity and identity for AI agents.

Before OpenUK, Amanda spent 25 years as a technology lawyer. She was the first lawyer at Canonical, the company behind the Ubuntu operating system, where she set up its global legal function, and later held senior legal roles at Global Switch and VEON. She sits on the board of the Mojaloop Foundation, is an Honorary Professor at the University of Bristol’s Smart Internet Lab, and previously served on the Cabinet Office Open Standards Board and the British Standards Institution’s AI standards committee. She edited Open Source Law, Policy and Practice and has been named one of the 50 most influential women in UK tech.

COVERAGE

– What the people building AI agents are focused on, and what they are finding harder than expected
– What open should mean when an AI model is described as open
– Who should be responsible when an agent gets something wrong
– What real control over AI looks like, and where the sovereignty debate helps or gets in the way
– Why well-described data and open standards make agents more reliable, and what to have in place before an agent is allowed to act

AUDIENCE

Chief data officers and heads of data governance. CTOs, CIOs and AI programme leads deciding which agents to build or buy. Product and engineering leads building with AI models. Legal, procurement and risk leaders who sign off the models, tools and contracts that agents rely on. Policy teams in government, the public sector and multilaterals shaping how AI is bought and governed.

DISCLAIMER

Amanda appears in her capacity as CEO of OpenUK. Questions are framed around her own experience, what she has observed in conversations with the people building AI and with governments, and OpenUK’s published research. Organisations from those conversations are not named, and she does not speak for the boards and companies she advises.