AI: 10 proof points
Every row that fits this shape of role: the rows the facts page shows first, then the rest in content order. Tom's own account of it: AI. All shapes, the facts page.
| Role | Product | Situation | Task | Action | Result | Year | Skills |
|---|---|---|---|---|---|---|---|
| Associate Product Director at Novo Nordisk | GenAI proof of concept for automating compliance reviews Agentic RAG proof of concept for compliance reviews, with a Figma plugin | Every release of health software needed sign-off from medical, legal and regulatory reviewers. The process was built for marketing materials, not software: every screen was manually captured as a PDF and reviewed manually, stacking the regulatory risk to the end of the development cycle and sometimes adding months to delivery. | Test whether an LLM could take on the review so product teams could shift left: fail in design, not at review. Two acceptance tests set up front: useful to product teams, and catches what the manual review catches. | Used an internal LLM sandbox to build a RAG model of the guidelines and regulations, producing a full audit trail per submission. Built a prototype agent that plugged into Figma so designers could query the guidelines before investing effort. Pitched it to senior leadership & tested with product team. | Product team loved it. The proof of concept showed a clear path to much faster release cycles, but the timing was 12 months too early & the business was not ready for AI. Tom moved to help the Wegovy Pill launch. | Feb 2022 - Feb 2026 | AI, GenAI, regulated, compliance, pharma, healthcare, discovery, prototyping, ways of working, launch, marketing, team leadership, stakeholders |
| PM/GM at LEO Innovation Lab | Diagnosis App Prototype smartphone app diagnosing skin conditions with machine learning | A moonshot: could a smartphone camera diagnose skin conditions? A 12-person team of data scientists, ML and computer vision engineers, wired for logic and certainty, in the deep uncertainty of product discovery. His first deep experience leading an ML team. | Lead the product direction, prioritisation and strategy, and get a working prototype that could be evaluated. | Worked with dermatologists to redesign closed-loop diagnosis and the ML training platform. Bootstrapped the ML engine with clinical decision trees. Designed the pipeline for data collection and analysis. Models were evaluated against a set of pre-tested images the dermatologists had all voted on. Not a medical device by design: disclaimers made clear it was not a diagnosis. Evaluated go-to-market and identified seven paths to market. | Over 80% accuracy across three skin conditions. Doubled the labelling output of the in-house image labelling platform. | Nov 2018 - Jan 2020 | AI, machine learning, computer vision, healthcare, discovery, prototyping, growth, launch, team leadership |
| Senior Product Manager at Novo Nordisk | Wegovycare Wegovycare weight-loss prediction model | Patient research showed everyone wanted the same three things: how much weight they will lose, whether this is normal, whether they are on track. Nothing in the brief answered them. | Make those three questions the core loop of the product, and find the growth lever inside them. | Built a weight-loss prediction model answering the first question: a predicted trajectory for the individual to check progress against. Started as an Excel spreadsheet on a laptop; the maths worked out with the strongest backend engineer, then taken from concept to prototype. The same check-ins fed the profiling engine. | Patients could see whether they were on track against their own projection. The model was later spun out as a Bayesian API service, a product in its own right, and the team extended it from weight to HbA1c. | Mar 2020 - Feb 2022 | algorithm, API, regulated, pharma, healthcare, patients, discovery, prototyping, ways of working, growth, retention, roadmap, 0 to 1, marketing, team leadership, stakeholders, consumer |
| Senior Product Manager at Novo Nordisk | Wegovycare Wegovycare profiling and customisation engine, delivered as an interactive magazine | Research showed every patient journey is different and every patient has different needs. A linear app would not work, and the drop-off risks were different for each person. | Identify each patient's drop-off risks and act on them automatically, with a system that coped with real people who do not check in on schedule. | Tom's own idea and pattern, most of the algorithm his. Built a continuous profiling engine combining clinical data, usage signals and PROMs. Applied a consumer marketing playbook to act on the top risk: it decided what content was shown and could trigger a messaging flow. Ran on fuzzy logic, tolerating gaps and misuse. Output to the patient: an interactive magazine that changed with them. | Every patient got a personalised experience from one product and one codebase. Part of the launch that reached 10% adoption in the first market and became the fastest-scaling digital product at Novo. | Mar 2020 - Feb 2022 | algorithm, API, regulated, pharma, healthcare, patients, discovery, prototyping, ways of working, growth, retention, roadmap, launch, 0 to 1, marketing, team leadership, stakeholders, consumer |
| PM/GM at LEO Innovation Lab | BetterVisitBot a pre-briefing chatbot for doctor consultations | Patients sometimes get intimidated by the doctor, and don't get the most out of their consultation, or forget what was said. | Prototype it with GPs. | Hid a dermatologist in an adjacent room to simulate a consultation chatbot. | Right idea, too early. Now it isn't. | 2016 to 2020 | AI, machine learning, computer vision, healthcare, patients, discovery, prototyping, growth, launch, team leadership |
| Solo projects, building with AI | TomBot TomBot, a chatbot CV at takenbyninjas.com | Tom wanted to create a novel way of presenting his CV, and find out first-hand what AI can and cannot do. Accuracy of recall was the key criterion: a CV bot that hallucinates is worse than no bot. | Build something real with AI and ship it, in his own content and voice. | Built with Claude Code on the Claude API: chunked RAG vector store, eval suite, guard-rails. Question-to-question matching against a bank of pre-mapped questions rather than searching raw content: it limits the chatbot nature but keeps the answers accurate. Content and tone of voice his own. | Live January 2026. Linked from every CV with a UTM tag per employer so each application's traffic is visible. Pinned on LinkedIn. | 2026 | AI, GenAI, regulated, compliance, prototyping, team leadership |
| Solo projects, building with AI | kittenpm.com kittenpm.com, a free browser-based product management training SIM inspired by Lemmings | Wanted a portfolio piece that is not a portfolio, aimed at product peers and recruiters. | Generate reach and demonstrate product skills by teaching product trade-offs as a game. | Built with React, TypeScript and Phaser, using Claude as the engineering team. Players practise balancing product trade-offs in real time; those who complete the game get a digital certificate to embed on LinkedIn, which drives further reach. | Live July 2026, in public beta. | 2026 | AI, GenAI, regulated, compliance, prototyping, team leadership |
| Solo projects, building with AI | deathspiralorgoldenage.org deathspiralorgoldenage.org, a weekly public poll, live sentiment tracker and open economic model on AI's economic impact, with CC0 data | A thought experiment that spun out from a discussion at a Claude meetup with Anthropic. | Build a public, open-data way to track the question over time. | Researched and built with AI. Content is human; the site is engineered and researched with AI. Weekly poll, sentiment tracker and an open economic model, all data released CC0. | Live September 2026. | 2026 | AI, GenAI, regulated, compliance, prototyping, experimentation, team leadership |
| PM/GM at LEO Innovation Lab | Diagnosis App data pipeline Image labelling platform and data pipeline for the diagnosis app | Labelling by dermatologists was critical for model performance, and a significant bottleneck. | Get more labelled data through the system without lowering its quality. | Interviewed every dermatologist on the labelling work to understand how they actually worked, then redesigned the in-house image labelling platform around that. Designed the pipeline for data collection and analysis to feed it. | Doubled dermatologist labelling throughput. Less of their time on labelling training data, more labelled data for the models. | Nov 2018 - Jan 2020 | AI, machine learning, computer vision, healthcare, discovery, prototyping, growth, launch, team leadership |
| PM/GM at LEO Innovation Lab | Diagnosis App paths to market Go-to-market evaluation for a moonshot ML prototype in dermatology | A functioning prototype, but heavy regulatory barriers and no clear path to market. | Work out how, or whether, a moonshot prototype could reach a market. | Evaluated go-to-market systematically and mapped the options with the team. | Identified seven paths to market, with the trade-offs of each, for the lab to decide on. | 2019 | AI, machine learning, computer vision, regulated, healthcare, discovery, prototyping, growth, launch, team leadership |