Writing
Blog
Thinking about AI product leadership, building with AI tools, governance, and the economics of shipping AI in production.
94% Capable. 33% Deployed. The Gap That Explains Everything.
Anthropic's new research reveals a 61-point gap between AI capability and actual deployment. That gap explains why the workforce apocalypse hasn't arrived, and what happens when it closes.
I Don't Deal in Hype: Why AI Product Leaders Must Also Build
Product leaders who haven't felt latency or been frustrated by hallucinations build AI strategies on fantasy. The case for a builder-leader identity.
Your AI Budget Line Is Wrong. Tokens Are the New Headcount.
Engineering teams are spending more on AI tokens than junior salaries. The cost structure of building software has inverted and most finance teams haven't noticed.
Two Production SaaS Platforms, One Builder: Solo Vertical SaaS
50+ AI features, 6 LLMs, native mobile, Stripe billing. Two vertical SaaS platforms built solo to validate the model transfers.
The E-Shaped Product Leader: Why Stacking Skills Beats Specialisation
AI collapses the cost of cross-domain competence. The career advantage belongs to people who stack skills, not the ones who go deeper in a single silo.
The AI Pricing Stack: Usage, Outcome, and Hybrid Models
Per-seat pricing is dying but the replacement isn't simple. A framework for AI pricing: usage-based, outcome-based, and hybrid models.
Constraint Is the AI Adoption Strategy Your Org Won't Try
Unlimited headcount kills AI adoption. One engineer per project, unlimited tokens, and the constraint to figure it out produces the best AI-native work.
AI Governance Without the Bureaucracy: A Risk-Tiered Framework That Ships
Most AI governance is either theatre or a bottleneck. A risk-tiered framework from shipping AI to AFSL-regulated Tier 1 banks.
AI Didn't Kill Coding. It Killed Typing.
AI coding is the sixth abstraction layer in 80 years: calculators, machine code, assembly, compiled languages, scripting, and now AI. Each layer was dismissed by the one below.
How to Measure an AI Product (When Traditional Metrics Lie)
DAU, time-in-app, and NPS were built for a world where humans do the work. AI products need different metrics. A framework for what to measure and why.
The Bitter Lesson Kills Your Orchestration Layer
Scaffolding gives you 10-20% gains that the next model wipes out. The bitter lesson for product builders: give the model tools and a goal, not a workflow.
Stop Reading About AI. Start Shipping With It.
The window for AI-enabled builders to capture outsized value is open but narrowing. Vertical SaaS will be built by individuals, not armies.
Enterprise AI Adoption Playbook: Why Seats Go Unused
AI features that work in demos fail in deployment because adoption is a product problem, not a training problem. A playbook from rolling out AI to Tier 1 banks.
The Agent That Won Didn't Have a Better Model. It Just Got Shit Done.
Weekend build to 145K GitHub stars to acquisition in weeks. The lesson: agents that execute locally instead of chatting in a browser win on adoption.
Build for the Model That Doesn't Exist Yet
Your AI product's market fit depends on a model that hasn't shipped. Build for the capability curve, not the capability snapshot.
Voice Agent Playbook: AI Phone Calls in Production
I built AI voice receptionists that handle real phone calls for real businesses. Latency, conversation flow, graceful handoff. Here's what actually matters.
The Cartel Problem: Why AI Stalls at the Industry Gate
AI's biggest obstacles aren't technical. They're structural: professional guilds, regulatory capture, procurement bureaucracy, and incumbents who profit from friction.
Your RAG Pipeline Is a Product Decision, Not an Engineering One
Chunking, retrieval, and grounding aren't engineering details. They're product decisions that determine if your AI helps or hallucinates.
Latent Demand Is Your AI Roadmap
The best AI products aren't imagined. They're discovered by watching how people misuse your existing ones. A framework for finding what to build next.
Product Discovery Is Dead: Why Prototyping Replaced It
The 6-week discovery sprint is a relic. When you can build a working prototype in a weekend, the fastest path to insight is shipping, not researching.
The AI Usage Gap Is a Product Architecture Problem
Most AI tools are deployed but unused. The friction isn't capability. AI lives in a separate tab instead of where work happens. Build inline, not destination.
Your Killer Feature Will Be Cloned by Friday. Build a Moat That Can't Be.
4 engineers, 10 days, a new product line. AI coding agents collapsed build economics. If code is a commodity, your moat is data, integrations, and trust.
Taste Is the Last Skill AI Can't Commoditise
AI commoditises execution. The scarce resource is knowing what to build, for whom, and when to stop. That's taste, and it's the career bet worth making.
Your Agent Evals Are Vibes. Here's How to Make Them Infrastructure.
Most teams evaluate agents with manual chats and gut feel. A practical framework for eval suites that let you ship, starting with 20 examples, not 20,000.
The Translation Layer Is Dead. Here's What Replaces It.
Agentic coding compressed the PM translation layer to zero. The three skills that matter now: problem shaping, context curation, and taste.
AI Strategy Needs Hands-On Experience, Not Slide Decks
AI strategies fail because leaders set direction for a capability they've never used. You can't strategise for a material you haven't touched.
Your Job Isn't Going Anywhere. Your Tasks Are.
AI doesn't replace jobs. It replaces tasks. The distinction changes everything about how you plan your career and your org chart.
The Cannibalisation Paradox: Why Per-Seat Pricing Dies in the Agentic Era
If your AI roadmap succeeds, your customers need fewer seats and you earn less revenue. The fix is pricing built around work units, not logins.
The 2,500% Audit Tax: The Math That Will Kill Your Multi-Agent P&L
A manager model checking every worker output increases unit cost by 2,500%. The fix is a spot-check architecture that saves 75% of your margin.
The Hard Hat Era: Why Your 2026 AI Strategy Is an Org Chart, Not a Chatbot
The shift is from prompt engineering to designing multi-agent hierarchies: AI managers overseeing AI workers that operate invisibly in the background.
Your AI Copilot Is a Margin Trap. Build for Replacement, Not Assistance.
AI isn't a feature. It's a new compute paradigm. Bolting GenAI onto legacy platforms destroys unit economics. If the AI is optional, it's a gimmick.
Text Is a Terrible Interface for Doing Business. Generative UI Is the Fix.
Google's A2UI signals the end of the chatbot text wall. Agents that render UI instead of paragraphs change what product teams actually build.
Stop Building AI Agents. Start Building SOPs Wrapped in Code.
A 5-step agent at 95% accuracy per step is only 77% reliable. The path forward isn't better agents, it's narrower ones. Three rules for workflows that ship.
Google's Real Estate Listings: Why Aggregators Should Worry
Google is testing native property listings in search. AI broke the constraint that stopped them in 2011. The aggregator model dies when the user is an agent.
The Billion-Dollar Team of One: Three Questions That Expose Your Org Chart
AI is breaking the link between revenue growth and headcount growth. Three questions that expose whether your org chart is designed for 2019 or 2026.
Stop Picking Winners in the Model Race. Build the Router Instead.
Building for a single model is technical debt. The winning strategy is orchestration, evals, and governance, not leaderboard loyalty.
The 17x Advantage: OpenAI's Product Builder Gap
OpenAI's enterprise data shows the top 5% of AI users send 17x more coding requests than the median. Three stats that redefine team enablement.
The Product Manager Is Dead. Long Live the Product Builder.
Meta PMs vibe code prototypes for Zuckerberg. LinkedIn scrapped their APM program. The PM role is being redefined, and the new skillset is prompt, build, eval.
3 Agentic AI Patterns from Google's Playbook
Strip the vendor marketing from Google's AI Agent Handbook and three stack-agnostic architectural patterns emerge that every product builder should steal.