davidfacer.com / aimaturitymodels.com / AI-Native Maturity Models / AI-Native SDLC / D9
Environment management
Ask an agent (or a new hire) to spin up a working copy of the system and see how long it actually takes, and how much of that time is someone manually configuring something that should have been reproducible. This dimension measures how far an organization has moved from "environments are hand-maintained snowflakes" to "environments are disposable, on-demand, and nobody has to remember how to build one."
Where most organizations start (Nascent)
No standard environment separation exists — development happens against a single shared, mutable environment, or directly against production. The first real step is establishing standard environment classes (dev, QA, staging, production) with documented purpose, access, ownership, and a defined promotion path, even if provisioning itself is still manual.
Where the real gains happen (Modeled → Integral)
The meaningful shift is defining environments as code — capturing infrastructure, configuration, dependencies, and policy in versioned, repeatable definitions, so rebuilding an environment from the definition produces something materially identical, no undocumented manual configuration required. From there, the real gain is on-demand provisioning: a pull request, command, or approved workflow spins up a fit-for-purpose environment and tears it down predictably, starting with whichever environments benefit most from ephemerality.
What the top of the curve actually looks like (Telemetric)
At full maturity, agents provision and tear down environments themselves, within explicit policy, quota, and security boundaries — determining when an environment is needed, instantiating the right template, verifying it's ready, and removing it when the work is done, with exceptions and failures still visible to human operators. Environment lifecycle stops requiring standing human maintenance.
Worth knowing about this ceiling: this model's own review checked Level E's "no standing human maintenance" bar against current agentic-infrastructure practice (sub-second provisioning, rollback-to-checkpoint as the default recovery pattern) and confirmed it holds. David's own review: "Correct as stated. The ideal is that architecture and guardrails are sufficient without standing human maintenance."
Why this dimension matters
Environment management is quiet infrastructure — nobody notices it until it's the thing slowing every other dimension down. An organization can be excellent at generation, verification, and release, and still lose most of its velocity to environments that take an hour to stand up correctly.
Drafted from the SDLC model's real locked content — including the transition and verification notes now folded into ai_native_sdlc_maturity_model.md itself (2026-07-27). D9's Level E ceiling question, named in this model's own Open Items, was reviewed and resolved 2026-07-24 — no open review flag remains.
Drafted from the SDLC model’s real locked content.