Product × AI

Where intelligence
meets execution.

AI isn't a feature — it's a fundamental shift in how products are conceived, built, and experienced. Here's how I think about the intersection of product strategy and artificial intelligence.

Perspectives

Six ways AI is
changing the craft.

🧠
AI as a Product Thinking Partner
AI compresses discovery cycles — from user research synthesis to hypothesis generation. PMs who treat AI as a co-thinker will outpace those who use it only as a writing tool.
In practice:A 12-person healthtech startup halved their user research synthesis time by running interview transcripts through a structured AI summarisation workflow — freeing the PM to spend time on synthesis rather than transcription.
⚡
Faster Iteration, Higher Stakes
AI accelerates build velocity dramatically — but that makes strategy more critical, not less. Speed without direction just gets you to the wrong place faster. Judgment is the new differentiator.
In practice:An enterprise ITSM team shipped 4× more A/B tests per quarter after adopting AI-assisted logic generation — but attribution errors multiplied until they added a mandatory strategy review gate before each test.
🔮
From Features to Intelligent Systems
The next generation of products won't have feature lists — they'll have capabilities that adapt, learn, and anticipate. Product strategy must evolve from shipping features to designing intelligent behaviour.
In practice:A mid-market analytics platform replaced its static report builder with an AI-generated narrative layer — adoption rose 34% because users stopped needing to know which chart to pick.
🎯
The PM's Role in an AI-First World
The PM is no longer just the voice of the customer — they are the architect of how intelligence is applied. That means owning the model selection, the guardrails, the explainability, and the trust layer.
In practice:A 3,000-person cybersecurity platform discovered their PM had no visibility into the model's confidence thresholds — leading to customer-facing errors that support, not product, was catching first.
🔗
AI Governance Is a Product Problem
Bias, hallucination, data residency, audit trails — these are not engineering afterthoughts. They are product decisions that determine whether enterprise customers buy, deploy, and trust what you ship.
In practice:A pharma tech vendor lost a $2M enterprise deal when procurement discovered their AI module had no audit trail — a product decision that was never made, not a technical limitation.
📐
Design for Human-in-the-Loop
The most durable AI products are not fully automated — they amplify human judgment rather than replace it. Designing the right handoff points between AI and human is the craft skill of this decade.
In practice:A mid-market HR platform reduced manager friction by 60% after redesigning their AI compensation tool to show reasoning alongside recommendations — managers trusted it once they could override it with confidence.
Principles

How I build
AI into products.

01
Start with the decision, not the data

The best AI product questions aren't 'what can we do with this model?' but 'what decision does the user need to make better?' Work backwards from the decision.

02
Explainability is a feature, not a footnote

Enterprise users don't trust black boxes. If your AI can't show its reasoning, it won't be used. Invest in explainability from day one, not as an afterthought.

03
Measure outcomes, not outputs

An AI that generates more recommendations is not better than one that generates fewer, better ones. Define success by downstream human outcomes, not model metrics.

04
Fail safely, learn systematically

AI systems will be wrong. The question is: how wrong, how often, and with what consequences? Design failure modes before you design success states.

Craft Skills

The craft skills that
matter most right now.

AI products fail in organizations not because the models are weak — but because the product decisions around them are wrong. Governance is a product problem. Explainability is a feature. Human-in-the-loop design is the craft skill of this decade. Whether you are a 20-person startup or a 10,000-person enterprise, these decisions determine whether AI gets used or gets abandoned. These aren't principles I read about. They're lessons from shipping AI in regulated, high-stakes environments at scale.

Coming Soon

Essays & frameworks
in progress.

Coming Soon
Why Your First AI Hire Is the Wrong First Move

The change management failure organizations make at every stage before they realize it.

Coming Soon
The 30-Day AI Foundation Checklist

What to fix before you write a single line of AI code, regardless of company size.

Coming Soon
Build vs. Buy: The Framework Organizations Get Wrong

How to make the call without wasting a year or a budget.