Brownfield development with Codex
Point of view
Codex is most valuable when it helps a team recover the system map, define the precise delta and verify that important existing behavior remains intact.
Brownfield change path
Recover the map
Identify entry points, boundaries, dependencies, data flows, conventions, tests, deployment paths and incident history. Ask Codex to explain evidence and cite the files or artifacts behind its model.
A codebase map is more useful than a file-by-file summary because risk travels through relationships.
Lock the present
Characterization tests and operational baselines establish what the system currently does—even when that behavior is imperfect. The new specification should state exactly what changes and what must not.
Small, reviewable increments reduce the distance between a change and its explanation.
Plan migration and return
Data conversion, compatibility, staged rollout, observability and rollback belong in the implementation plan. Codex can accelerate analysis and scaffolding, but the team owns the transition.
Treat surprising legacy behavior as information before treating it as debris.
Design principles
What changes Monday
Ask Codex for an evidence-cited system map.
Choose one focused seam and add characterization tests.
Write the change and the non-change contract.
Define migration, canary and rollback before implementation.
The goal is to improve a mature system while preserving the qualities and business knowledge that made it last.
Discuss this perspectiveContinue exploring
AI can accelerate delivery only when intent, boundaries, evidence and failure behavior are explicit enough to govern the work.
Read field note
An AI-native lifecycle changes the system of work around evidence, review gates and accountability.
Read perspective
Generated tests create value when they challenge behavior, boundaries and failure using an independent definition of correctness.
Read field note