Insight · Governance
The rules our delivery agents work under
Four standing rules, how each is enforced, and the drafting practices that sit beside them. Written for a CTO and the security reviewer who has to sign off.
Insights
Practical thinking on AI-led delivery
Articles from the people who build and run Anitech's products and services.
Articles about Governance
Insight · Governance
An override rate of zero is an alarm
Autonomy should be earned one step class at a time, and withdrawn automatically. Why a reviewer who never overrules the agent may have stopped reading.
Insight · Measurement
Why we set the pass mark before we ran the test
We built an estimator for how long new requirements take, wrote down the result it had to reach, and it fell short. So we didn't ship it. Here is what that tells you about evidence in AI-led delivery.
Insight · Harnesses
What a two-person pod did in two days, and what it cost to get there
49 user stories to opened merge requests in about 48 elapsed hours, after seven working days of setup. The result, what it does and doesn't show, and what we didn't measure.
Insight · Measurement
Where delivery time goes: nearly half of it is waiting
In 587 timed stories from one client's delivery history (of 1,244 in total), about 47% of story time was waiting, and 36% of stories re-entered a status. Why faster coding alone doesn't help.
Want to talk it through?
Tell us what you are working on and we will say how we would approach it.