aiXP

Appearance

”AI Extreme Programming"Extreme Programming for Human-AI Teams of 1 to N

Working draft - brainstorm stage.

Start reading
Bibliography

Premise

Extreme Programming was built on one bet - tight feedback loops beat upfront certainty. AI makes that bet stronger, not weaker, and shifts the cost of software from writing code to verifying, bounding, and integrating it.

The Practices

From Spec-First Prompting and Asymmetric Pairing to Agent Role Topology and Risk-Tiered Verification - engineering practices designed for machine-speed collaboration.

The Trust Factory

Grounded in classic XP, Lean, and SRE. Software is bipedal - code and trust must advance together.

A note on intent ​

A recurring thread in developing this has been how few working engineers seem to have a coherent model of practices like regression suite maintenance, tiered release strategy, or versioning discipline - not because these are obscure, but because they're rarely taught as a connected system, and AI tooling is what now makes running that full system affordable at small scale. Part of the goal here isn't just documenting one practitioner's method - it's making practices that used to require org-scale headcount teachable to individuals and small teams who now have the tooling to actually run them.