Disciplines → 1 of 6

AI-Augmented Delivery

AI in every stage of the lifecycle, from business concept to retirement, inside the delivery practices you already run.

Overview

AI-Augmented Delivery is where AI meets the work itself. It covers how AI takes part in each stage of the lifecycle, from shaping a business case to decommissioning a service, and how the pipeline proves that every AI-assisted change was checked before it shipped.

The discipline does not introduce a separate "AI process". It adds AI contributions and matching controls to the stages teams already run, so adoption changes how work is done without changing where it is recorded.

  1. Conceivebusiness case
  2. Planbacklog
  3. Specifyspecs, ADRs
  4. Generatecode, IaC
  5. Verifyreview, evals
  6. Releasetiered gates
  7. Operaterun, respond
  8. Learnadjust
  9. Retiredecommission

What changes in existing practice

PracticeTodayWith AI-Augmented Delivery
RefinementStories written by handAI drafts stories and acceptance criteria; the product owner owns acceptance
DesignADRs record a decisionAI proposes options; the ADR records AI involvement and the human decision
DevelopmentEngineer writes the changeAI generates code, tests and IaC; the engineer owns intent and correctness
CI/CDBuild, test, deployAdds provenance, licence and eval checks, and approvals driven by risk tier
DecommissioningServers and servicesAlso prompts, agents, integrations and non-human credentials

Key practices

Tier the work before it starts

Assign a risk tier when an epic or change is planned, not when it reaches review. The tier decides how much AI may contribute and which approvals the pipeline will demand, so engineers know the rules before they generate anything.

Specify before you generate

Generated code is only as good as the specification it answers. Acceptance criteria, interface contracts and test cases are agreed first, often drafted by AI and approved by a person. Verification then checks the change against an agreed intent, not against whatever the model produced.

Mark every AI-assisted change

Record AI involvement where the work is already recorded: a vendor-neutral commit trailer such as AI-assisted: yes, a field on the pull request, a line in the ADR. The audit trail then comes from the version control system and tracker, with no separate store to maintain.

Make the pipeline AI-aware

Keep the existing stages and add controls: secret scanning and context hygiene before commit; licence and IP checks on generated code; evals for prompts, agents and AI features; and a policy gate that maps the change's tier to the approvals required. The reference pipeline shows where each control sits.

Keep an element register

Every element created during delivery has an owner, a tier, a provenance record and a state from proposed to retired. Start with what you already track, such as code, releases and access, and add the AI-specific elements: prompts, context packs, agents, tool integrations and non-human identities. See the element register.

Retire deliberately

AI adoption creates assets that outlive their purpose: prompts nobody owns, agents with standing credentials, integrations to tools no longer approved. Retirement is a lifecycle stage. It finds dependants, revokes access and archives provenance, so nothing keeps running without an owner.

Controls by risk tier

TierPrevent (Guardrails)Prove (Audit Trail)Detect (Self-Monitoring)
LowApproved tools only; automated checks must passProvenance trailer on the changePipeline failure rate on AI-assisted changes
MediumOne qualified human reviewer; licence and secret checksPull request links spec, review and provenanceChange failure and rework rate, AI-assisted vs not
HighTwo reviewers incl. a specialist; evals or security tests mandatoryEvidence bundle attached to the release recordEnhanced monitoring for a defined window after release
CriticalHuman-only authorship and decision; AI may analyse, never actDocumented rationale; second-line sign-offAny AI involvement triggers review

Maturity path

  1. Ad hoc — engineers use AI tools privately; no one can say which changes were AI-assisted.
  2. Experimenting — named teams pilot approved tools in development; results shared informally.
  3. Managed — tiers applied at planning; AI-assisted changes marked; review depth follows the tier.
  4. Governed — the pipeline enforces provenance, licence, eval and approval gates as policy-as-code across all stages.
  5. Optimised — delegation per stage adjusted from change-failure and rework data; retirement automated from the element register.

Assess your organisation →

Measures

Anti-patterns

Regulatory anchors

For organisations deploying AI tools, the EU AI Act's AI literacy duty (Art. 4) applies to everyone using them in delivery. ISO/IEC 42001 expects documented information on AI use (clause 7.5) and operational control (clause 8); marking changes and keeping the element register provide both from the systems already in use.