Five levels of maturity. Each one unlocks when the layer below has proved itself. The pace is set by your risk appetite and the maturity of the standard — not by a calendar.
Each level is something concrete the team can point to. No theoretical phases, no "transformation roadmaps" — a working system at every step.
Narrow customer-facing apps. The system is mature enough to expose externally — clients pay for the outcome, not the hours.
The system is reliable enough that your team depends on it day-to-day. The data moat compounds with every engagement.
Validated patterns become plugins, skills, SDKs, or no-code tools your team uses every day. Cycles of testing, then locked in.
Active testing. Real results land — but consistency and reliability are missing. The team finds what's worth keeping.
Your team uses Claude, GPT, or Gemini comfortably. Working with agents becomes everyday behaviour. The foundation everything else stands on.
Internal first, external second. The system earns its keep inside the business — capacity released, EBITDA per head lifted — before it ever sees a customer. When it does, it's because the standard is mature enough that exposure is a release decision, not a leap of faith.
The shift from services to software isn't about "adopting AI." It's about separating what only humans should do from what humans shouldn't have to do, then building the system that runs everything else — in a way that outlasts our involvement.
The intelligence layer becomes software. The judgement layer — relationships, taste, trust — stays human. Your people work on what only people can do.
Most strategy firms stop at the slide deck. We don't. The value lands when we're shipping fixes with your team — we run the system alongside them or build it for you.
AI is commodity. Defensibility comes from the proprietary data your firm has been generating from delivery for years — campaigns, claims, contracts, calls. We turn it into the data your AI runs on.
Pick one painful task. Ship a fix in a fortnight. Let your team experiment — break it, retune it, find what works in real use. When the pattern holds, we lock it in. Then pick the next. Strategy emerges from what's working — not the other way around.
Your services book stays services and trades at services multiples. The new software-revenue line trades at a software multiple. The blended exit lifts — and the story survives diligence.
Standardise so every output meets the same bar — every client, every campaign, every time. Documentation, runbooks, capability transfer build the system that holds it. The team gains time; the work holds its quality.
A 30-minute conversation to map your service lines, surface the first thing worth automating, and figure out which level of the stack you're already on.