AI hasn’t removed engineering work. It’s changed where the leverage is.

Writing code is getting cheap.
Trust is getting expensive.

So my day-to-day is shifting up the stack:

Turning intent into a spec (goal + acceptance criteria)
Setting constraints + guardrails (what we will / won’t do)
Insisting on verification (tests + CI as the gate)
Making changes auditable (run logs, evidence, rollback paths)
Breaking work into small, reversible slices

I’ve built an AI “dark factory” to support this with a playbook based on the production practices I’ve relied on for ~20 years (TDD/verification, refactoring moves, design patterns, sensible boundaries).

The goal is simple: PRs should arrive already at the standard and prove it.

So the workflow is:

AI ships changes as PRs with:
• a scoped prompt/spec
• a run log (commands + test output)
• CI green as the gate
• evidence attached next to the diff

If it’s green, it auto-merges.
If it’s not green, it troubleshoots + iterates until it is.

Review stops being “read 500 lines and hope”
and becomes “check the contract + verify the proof”.

That’s what I mean by “software factory”.
And that’s what I’m managing.