Skip to content

New here

Start here

I write about building software with an AI agent fleet, the product judgment that decides what gets built, and what a small market teaches you about both. Here are the paths through it.

Start with these

If you read three things

The arguments the rest of the archive builds on — the system I work in, the discipline that makes it safe, and the market that shaped both.

  1. My platforms are built almost entirely by AI agents. Here's the system.J is a multi-agent suite my brother and I built: planners, implementers, a review panel shipping production code. The hard part isn't generation, it's process.2 min
  2. Supervision is the new managementWhen agents do the work, the manager's job doesn't disappear — it moves, from assigning and checking to policy and exceptions. Management theory saw it coming.2 min
  3. Building in a small market is a feature, not a bugRomania will never be anyone's TAM slide. That constraint shaped how I build products — and I've come to see it as an advantage, not a handicap.3 min

Track

Building with an agent fleet

What it actually looks like when most production code is planned, written and reviewed by agents — including the parts that stay human.

  1. We rebuilt our agent suite into a product team. Here's what the evidence said.J could take a ticket to a pull request but not decide what to build. The research is blunt: org charts lose, one writer with independent verifiers wins.5 min
  2. What AI agents changed about estimationEstimates used to be expensive guesses that aged badly. When a pre-flight agent sizes every ticket in minutes, the economics flip — and the arguments change.2 min

Track

Product judgment and delivery

Ten years of scoping, sequencing and killing work — the decisions no tool makes for you.

  1. The one-page scopeEvery project I've seen go badly went badly in the same place: the gap between what the client meant and what got built. One page closes it — exactly one.4 min
  2. Run your roadmap like a research programDoing a PhD while shipping products taught me an uncomfortable truth: most roadmaps would not survive peer review. Here is what research method fixes.3 min
  3. Why software teams miss deadlines: notes from research and practiceMy PhD studies how software teams deliver. The most cited finding is also the least glamorous: methodology matters less than the discipline around it.2 min
  4. From 6 to 13 squads: what actually breaks when you scale deliveryI scaled Toptal's Growth Product organization from 6 to 13 agile squads and served as interim VP. The bottleneck was never the number of engineers.2 min

Track

Building for a small market

Romania as a design constraint: regulation as a launch window, verification as a product, pricing outside the dollar zone.

  1. Compliance is a feature, not a checkboxIn regulated markets, the compliance calendar is a roadmap input. The teams that treat mandates as launch windows win entire adoption waves for free.2 min
  2. Shipping on top of government APIs: lessons from e-Factura and five state registriesBONO integrates with Romania's e-invoicing system; Firmoscop reads five official registries. Building on government infrastructure is a discipline of its own.2 min
  3. Why I built FirmoscopIn Romania, “should I give this company payment terms?” means an hour across five public registries. Firmoscop makes it one verdict: green, yellow, or red.2 min
  4. From marketplaces to fintech: trust is the productToptal taught me marketplaces sell trust with a matching engine attached. Fintech taught me the same with money attached. Everything I build now starts there.2 min
  5. Own everything: the vendor-replaceability testOne question exposes any technology relationship: if we stopped working tomorrow, what would you lose? Why I structure everything around a boring answer.2 min

Track

The long game

Career, craft, and the habits that keep working when the tools change every quarter.

  1. The product skills that actually compoundMost skills a PM learns depreciate — tools change, frameworks rotate. A few appreciate with every year and every job. Optimize for those.2 min
  2. Stay a student: keeping an open mind, operationalizedA learning mindset isn't a trait you have or don't. It's a system — forcing functions that make you update your beliefs whether you feel like it or not.3 min
  3. Thirteen certifications later: what credentials are actually forPMP, PMI-ACP, Six Sigma Master Black Belt, SAFe, PROSCI, Harvard CORe — more credentials than most people think reasonable. An honest accounting of their worth.2 min
  4. Doing a PhD while shipping productsDoctoral research on software delivery alongside full-time product leadership since 2018. People assume the two compete for time. Mostly, they feed each other.2 min
  5. The hard stopTen months ago I became a father. It didn't make me a better builder — it audited the systems I had, and some were nothing more than me staying up late.4 min

Series · Saturdays

The Weekend Read

Where AI and the tech industry actually moved this week — one argument, the signals behind it, and what's worth your time.

The system

How I build with AI

The fleet, the pipeline from ticket to production, and the rules it cannot break — the operating system behind most of these essays.

Get the next essay

Product, growth, and AI-assisted engineering — straight to your inbox, once in a while. No spam, unsubscribe anytime.