Notes · 13 articles
Field notes
What I’ve learned shipping product and AI systems to production: what blocks, what can be measured, what lasts. Short, concrete notes, written for teams who have to decide and deliver.
AI in production
From proof of concept to a system used every day: organisation, measurement, AI Act compliance.
- From AI POC to production: why it stalls, and how to get throughThe five blockers that hold proofs of concept back, and the sequence for moving to a measured production system.
- AI Act: classifying your AI systems (Annex III) in practiceInventory, reading Annex III, provider or deployer role, timeline: the classification method step by step.
- Measuring an AI support assistant: deflection, intent accuracy, resolution timeDefining, instrumenting and reading the three metrics of a support assistant, with the traps that inflate the numbers.
Product and roadmap
Product management fundamentals, updated for 2026: roadmap, backlog, MVP, market fit, hypotheses.
- Minimum viable product (MVP): how to design oneThe lean startup definition, choosing scope by segment, launch criteria and what AI changes about the method.
- Building an agile product roadmap in AsanaStructure, time axis, custom fields and the link to sprints: an agile product roadmap template ready to reuse.
- Product roadmap: how to write one (with an example)Who it is for, how to build it, an annotated example and the rules that keep it read.
- Product-market fit: definition, signals and methodA verifiable definition, the signals that matter, and a method built on niches and hypotheses.
- Product manager CV: what makes the differenceStructure, quantified results, ATS, the mistakes that get you screened out, and how to present credible AI experience.
- Agile roadmap: how product managers build oneThe planning cycle, what goes into an agile roadmap and the mistakes that drain it of meaning.
- RASCI: definition and an example matrix for a product teamThe five roles, how it differs from RACI, and two annotated matrices: a product team and an AI system in production.
- Product backlog: mistakes to avoid (and grooming that works)Content, prioritisation, useful grooming, and the seven mistakes that turn a backlog into a dumping ground.
- Agile roadmap: how to involve your teamsEight principles so developers and business teams build the roadmap instead of having it imposed on them.
- Product hypotheses: the roadmap that leads to market fitWriting a testable hypothesis, ordering by risk, deriving the roadmap and knowing when to pivot.