Architectures of trust
Notes toward a theory of why the best technical systems mirror the best social ones.
· Publications
The pieces I didn't rush.
Every publication, grouped by form. The same paywall + subscriber flow you see on the landing carries through.
Essays1
Articles10
When Speed Became Trust
Speed is a sensation. Trust is a judgment. The mistake of the last year has been using the first as a proxy for the second.
1 - Three Days, Three Months, and What I Had to Unlearn
The shift from a three-month to a three-day, end-to-end platform build is not a story of 30x productivity, but rather a re-shaping of work where AI removes constraints and renders traditional, slow development habits obsolete. By challenging assumptions like upfront architecture and long feedback loops, this change demands a new approach to software engineering where the process is no longer defined by old constraints.
9 - The Reckoning.
Most organizations think they’ve made the transition to AI-native when they’ve only deployed AI into the organization they already have. The real crossing begins after the productivity gains—when companies have to confront a harder question: if AI changes the cost of coordination, context, execution, and organizational memory, which parts of the operating model still make sense? That reckoning reaches far beyond technology. It touches roles, seniority, rituals, team structures, hiring, leadership, and ultimately the way organizations compete. AI-augmented and AI-native may look similar from a distance. The difference is structural.
2 - The Price of Agreement & Trust
For years, I believed careful upfront design was an engineering discipline. I now think much of it was something else: a coordination tax. I realized that the agreement between people was the expensive part. AI doesn’t eliminate that need, but it changes its price dramatically. Once I saw that, I started questioning which engineering rituals are still earning their cost or price.
8 - The AI-Native Organization
There’s a difference between an organization that uses AI and one that’s designed around AI. The first deploys new tools into an existing operating model. The second redesigns how work, coordination, memory, and decision-making actually happen. That’s the difference between being AI-augmented and being AI-native.
3 - The Shape, Not the Rate
We keep measuring AI by productivity gains because that’s easy to count. I think that’s causing us to miss the real transformation. Faster loops are interesting. Different loops change companies.
4 - Un-training Teams: Why AI Adoption Is Not Training
5 - The Wrong Question
6 - Organizations Built Around... Context?
