The AI SDLC
The software lifecycle re-implemented as agent-native infrastructure -- the pipeline spine the rest of the platform plugs into. A spec enters at one end; reviewed, tested software comes out the other, with agents moving the work through every stage and humans defining intent and approving at gates.
The Pattern
"DevOps gave us the SDLC. CDLC is the same discipline applied to the context that drives coding agents." -- Bo Fuller, 12-Factor AgentOps (source)
The AI SDLC is the software lifecycle re-implemented as agent-native infrastructure -- the pipeline spine the rest of the platform plugs into. Where the traditional SDLC moves human-written code through plan, build, test, review, deploy, and operate, the AI SDLC wires the same stages so that agents do the moving while humans define intent and approve at gates. Every familiar phase gets a counterpart: write the instruction, compile the context, validate the outcome, deliver to the agent, observe what happened, preserve the learning (12-Factor AgentOps).
Treated as platform, this is not a methodology slide -- it is a running system. It turns one developer's spec into a repeatable delivery line: the eval service is its quality gate, sandboxes are where construction runs, and observability traces the whole pass.
Why It Matters
As autonomous loops take over the build, the lifecycle compresses -- pioneering orgs describe sprints giving way to "bolts," delivery units where weeks become days, with humans setting intent and validating at stage gates (BCG Platinion). That makes the real bottleneck explicit: it is no longer coding speed but the clarity of intent and the reliability of verification. The honest caveat is readiness -- agents are only as good as the codified knowledge and pipelines they run on, and teams that skip clean repos and reliable CI/CD "risk automating chaos" (BCG Platinion). The AI SDLC is the investment that turns the workflow patterns into something a whole organization can run.
Sources
Last reviewed: 2026-06-24