Why an operating model
Agentic coding for the whole organisation. Not only solo devs.
Teams try Cursor, Claude, or Copilot on one stage at a time. Cool demos. No shared map of who owns what, where AI is allowed to draft, and which gate clears before the next lane starts.
An AI Operating Model makes that path explicit. Roles stay clear. AI accelerates the work around them. Releases move faster because everyone already knows the handoffs.
The delivery path
Five lanes. Overlapping time. AI marked where it actually works.
Process lanes are streams, not a waterfall. Calendar days can hold ideation, architecture, and build at once. Agent and assist chips show where AI does primary work versus where a human still leads.
Operating model · idea to release
Role lanes across time. Mint marks where agents lead. Dashed edges mean AI assists.
01
Ideation & Shaping
- Business defines the problem and success criteriaAssist
- Agent completes the intake artefactAgent
- Management validates intake and creates the epicAssist
- Architects clear the feasibility and routing gate
02
Architecture & UX
- Designers produce the UX artefacts
- Architects review the AI design draftAssist
- Architects sign off Ready for Development
03
AI Implementation
- Engineers write the failing tests first
- Agent implements against the testsAgent
- Engineers review and accept every AI diffAssist
- Agent runs the first-pass reviewAgent
04
Validation & QA
- Engineers complete the human PR reviewAssist
- Business accepts the release in UAT
- Management approves go-live
05
Deploy & Operate
- Engineers deploy to production
- Engineers capture monitoring and metricsAssist
- Management reviews CoE outcomesAssist
Principles
Speed with ownership, not speed instead of it.
01
Humans own intent
The spec, the test, the architecture decision, and the merge stay with people. AI accelerates those steps. It does not replace accountability.
02
Phases are streams
Ideation, design, build, and validation overlap on the calendar. Rework can reopen an earlier lane without pretending the path was a waterfall.
03
Roles change, they do not shrink
Business, architecture, design, engineering, and management keep their decisions. AI fills the repetitive drafting and checking around them.
How roles change
Each role keeps the decisions. AI fills the drafting around them.
Business
Keeps problem framing, acceptance criteria, uat. AI accelerates intake structure, clarifying questions, outcome summaries.
Also in Solutions
Need portfolio governance on top of the delivery path?
The AI Center of Excellence covers enablement, discovery, risk zones, and continuous monitoring across every AI-built app. The operating model is how engineering delivers inside that programme.
Start with your current path
Map AI into the lifecycle you already run.
A short session on where agents fit today, which gates need clarity, and what a shared operating model looks like for your teams.