How I think about building and leading engineering organizations in regulated industries — talent, governance, and organizational design for the agentic era.
Three convictions drive this operating model:
-
Talent density is the only metric that matters. AI amplifies existing capability. The output differential between exceptional and average engineers is no longer linear — it is exponential. The optimal engineering organization in the agentic era is smaller, more expensive per person, and more intolerant of mediocrity.
-
Attribution is a first-class design requirement. AI makes it easier to be a bad actor, faster. Every agent action needs a named human owner. Every pipeline decision needs a human decision it can be traced back to. Attribution is not bureaucracy — it is the condition that makes agentic capability trustworthy in a regulated environment.
-
Domain judgment is the new core engineering skill. Engineers become the new product managers — decomposing business problems into agent-executable work, verifying output, and making judgment calls at the boundaries where agents fail. The most valuable engineering skill in the agentic era is domain judgment, not technical execution.
- Organizational design: team topology, on-call model, agent management layer, governance structure
- The 90-day forcing function and 36-month transition
- The workforce question: what happens to engineers who are already here
- The organizational moat: why this model compounds and cannot be quickly copied
- Three non-negotiables that do not change under pressure
This document is the organizational companion to the technology strategy thesis:
| Repo | What it covers |
|---|---|
| platform-engineering-thesis | Why platform engineering is the prerequisite for AI in regulated industries |
| orbit-platform | The technical implementation of attribution-first governance |
| cab-automation | Automated change governance — the deployment gate in practice |
| fintech-platform-reference | Reference architecture for regulated platform engineering |
Creative Commons Attribution 4.0 International