01 · Maturity
Where has AI entered the SDLC?
Locate adoption across IDE, chat, pull requests, planning and custom agents, then distinguish personal assistance from workflow-level capability.
The AI-SDLC Transformation method
Make AI part of the delivery system, not another layer of tooling.
A practical method for engineering organisations that already use coding AI and now need to locate maturity, measure delivery impact, establish governance, and increase autonomy without losing accountability.
10,000+Engineers supported by the enterprise AI-SDLC transformation
14,000+Users moved to the GitHub platform foundation
Up to 3,000Engineers in a single enablement audience
15+ yearsSoftware delivery, platform, DevSecOps and cloud
One system, four decisions
Separate adoption, value, control and autonomy.
Vendor dashboards often collapse all four into usage. The transformation method keeps them separate so leadership can make an evidence-backed decision about what to scale, fix or stop.
02 · Measurement
What changed in delivery?
Connect usage to planning time, cycle time, review load, rework, quality, security and retained value per workflow.
03 · Governance
What keeps the system accountable?
Define ownership, data and policy boundaries, agent distribution, human approval gates and evidence that controls are operating.
04 · Autonomy
Which workflow can safely move next?
Choose one bounded workflow, instrument it, run it behind human gates, and raise autonomy only when its outcomes are defensible.
Seven-stage maturity ladder
Locate what is operational, not what is licensed.
The ladder is evidence-based. A stage is reached only when the relevant behaviour, workflow and controls can be observed.
Stage 1
Licensed
Seats and policy may exist. Daily engineering work has not materially changed.
Stage 2
Activated
A cohort has signed in and tried the assistant. Usage is uneven and mostly personal.
Stage 3
Regularly adopted
A meaningful share of engineers uses an assistant in weekly work, not as a novelty.
Stage 4
Multiple assistant surfaces
IDE, chat, pull-request and planning surfaces are in play, but context is still fragmented.
Stage 5
Embedded into workflows
AI is inside planning, implementation, review and documentation paths with shared instructions.
Stage 6
Agent-initiated with human gates
Specialist agents start work. Humans approve architecture, security and release decisions.
Stage 7
Increasing autonomy with measurable controls
Autonomy grows only where quality, security, value and intervention frequency are measured.
Eight measurement families
Measure the AI system, not adjacent operational noise.
No single metric proves value. The evidence has to connect adoption and behaviour to delivery outcomes, control performance and business value.
Licence activation and active usage
Who has access, who works with the assistant, and who has gone quiet.
Adoption by assistant surface
IDE, chat, pull-request review, planning and custom agents tracked separately.
User-initiated versus agentic execution
How much work starts with a person versus an agent opening or progressing the task.
Acceptance and completion
Suggestions accepted, tasks completed, and work abandoned after AI involvement.
Planning and cycle time
Time from intent to approved plan or from change to production, attributed at workflow level.
Quality and security
Defects, rework, policy violations and security findings on AI-touched work.
Autonomy and intervention
How often humans must stop, correct or approve an agent before work can progress.
Business value per workflow
What a governed workflow returns: compressed planning, cleared backlog, reduced risk or released capacity.
Controlled progression
Raise autonomy one governed workflow at a time.
The transformation does not start by deploying more agents. It starts by choosing a workflow where accountability, evidence and rollback can be designed before autonomy is increased.
Baseline
Find the bottleneck
Locate where AI is adding work, shifting review load, creating risk, or failing to change the delivery outcome.
Governed pilot
Instrument one workflow
Assign owners, define inputs and outputs, place human gates, collect the evidence and make rollback explicit.
Scale decision
Scale, fix or stop
Expand only when delivery, quality, security and intervention data justify the next autonomy level.
Principal-led, delivered through your engineering organisation. The method is designed to be operated by the team leads, champions and platform owners already inside the organisation—not by growing consultant headcount.
Start with evidence, then decide what to build.
The four-week Transformation Assessment applies this method to your organisation and ends with a board-ready decision pack, a governed pilot design, and a costed 30/90-day roadmap.