ProductTank Oxford

We're back with another in-person ProductTank Oxford meetup at the King's Arms in central Oxford. This month, we're delighted to feature two great speakers with talks on roadmapping and AI assurance.

6:00pm Doors open, drinks, and networking
6:50pm Welcome
7:00pm Phil Hornby - Going beyond the "Now-Next-Later" Roadmap
7:45pm Break
8:00pm Claire Woodcock - Nobody Should Trust Your AI, How to build systems people can rely on
8:45pm Wrap-up
8:50pm Close (but you're welcome to stick around!)

We hope you can join us!

Going beyond the "Now-Next-Later" Roadmap
Roadmaps are a visual tool used to communicate and align many audiences. Yet why do most of them just look so boring? Now-Next-Later is a great start in breaking bad patterns but there’s more to rethinking the roadmap. Great ones show goals, priority and optionality. They capture direction, discovery and delivery. Yet they stay simple enough for everyone to understand. Learn about 14 visual styles you can choose from to deliver the message and keep the audience interested. Let’s turn your roadmaps from dull documents into powerful tools.

Phil Hornby is a highly experienced product leader, coach, and technologist with over 20 years in the industry. His career spans various roles, including product management, project management, operations, development, sales, and marketing across diverse sectors like SaaS, IoT, and Automotive. Phil's mission is to help product leaders think clearly, make strong decisions, and take powerful action that drives high-impact outcomes. Phil has helped thousands of individuals across hundreds of companies, combining technical, business, and human skills to empower teams and leaders. His expertise lies in delivering complex solutions that elegantly solve customer problems at scale. His approach emphasises collaboration, context-driven leadership, and continuous learning.

Nobody Should Trust Your AI: How to build systems people can rely on
You can hardly move for people telling you to become AI-native. Vibe coding feels like magic. Your customer service team is suddenly handling twice as many queries. And yet, you can’t shake the nagging feeling that at any moment one of these systems might confidently let your product, your customers and ultimately, you, down.

You shouldn’t trust AI, and neither should anyone else.

Drawing on lessons from years of working in high stakes AI products, Claire will explore why hallucinations are only the obvious failure; how evals can reassure you without testing what actually matters; and how to build safeguards, monitoring and recovery paths that help you sleep at night.

Drawing on stories from real AI products, we’ll look at how to add AI without surrendering control, and build systems people can rely on without asking them simply to believe.

This is not a talk about email summarisation. Bring your own stories from the trenches to share.

Claire Woodcock is an AI product leader who has spent her career building products in domains where being confidently wrong matters. Most recently VP of Science & ML at Climate X, she previously led AI Products at Mozilla, Twitter, Onfido and Babylon Health across climate risk, recommendations, biometrics and healthcare. She has a postgrad in the Social Science of the Internet from the Oxford Internet Institute and has published research on explanations and trust in AI. Claire specialises in turning complex machine learning into useful, commercially successful products, whilst asking awkward questions about what they're good at.