Innovation economics
The ROI question everyone skips
Most AI business cases are built backwards, from a solution that already has a champion. Here is the version that survives a second budget cycle.
4 August 2026 · 4 min
AI strategy · Adoption · Return
AI strategy, roadmap and adoption - costed before you commit.
We help organisations choose the AI worth doing, prove it with a working prototype, and get their teams actually using it. Every recommendation carries the economics behind it.
Why AI programmes stall
The technology is almost never the reason.
Most stalled AI programmes share a post-mortem: a pilot that worked, a business case nobody could defend, and teams who were never brought along. The model was fine. The roadmap, the ownership and the change plan were missing.
We start from the other end - the use cases that pay for themselves, sequenced into a roadmap someone owns, with the training and change work that makes them stick.
Before enthusiasm
Every recommendation comes with a number.
We model the economics before anyone commits budget. If the number does not work, we say so - a cheaper outcome than discovering it in year two.
- Cost to run
- At your volumes, in production, including the runs that fail.
- Return
- Hours released, errors avoided, revenue enabled - measured against a baseline we agree first.
- Payback
- When the initiative stops costing money and starts returning it.
- What must be true
- The assumptions the case rests on, so you know which one to watch.
What we do
Six practices, one team
- 01
AI Strategy & Roadmap
Which AI is worth doing, in what order, owned by whom - and what it returns.
- 02
Adoption, Change & Enablement
The work that decides whether anyone actually uses what was built.
- 03
PoC & Rapid Prototyping
A working prototype in weeks, with an evaluation report and an honest verdict on whether to continue.
- 04
EU R&D Funding Strategy
Where genuinely novel work can be part-funded - and when a call is not worth entering.
- 05
Innovation & Product
Turning research results and prototypes into something with a price, a market and a margin.
- 06
Proposal Drafting & Technical Writing
Proposals, deliverables and architecture documents written by people who could also build the thing.
Where to start
Fixed scope, fixed price
Six ways in, each with a defined deliverable and an end date. Larger engagements usually begin as one of these.
2–3 weeks
AI Readiness & Opportunity Assessment
A scored view of where you actually stand across data, technology, people and governance - and the shortest path to the first result worth having.
2 weeks
AI Business Case & ROI Model
What the initiative costs to run at your volumes, what it returns, when it pays back, and what would have to be true for that to hold.
Monthly
Fractional CAIO
A Chief AI Officer on retainer. Architecture decisions, vendor selection, hiring input and roadmap ownership, without the executive hire.
4–8 weeks
AI Enablement Programme
Role-based training for executives, managers, engineers and analysts, plus the champions and policy needed to make it hold after we leave.
4 weeks
PoC Sprint
A working prototype against your real data, an evaluation report, and an honest recommendation on whether to continue.
3–5 days
Call Fit Check
A direct verdict on one specific call: your realistic win probability, what is missing from your position, and what entering would cost you.
An unusual advantage
Some of this can be paid for with European R&D funding.
Where the work is genuinely novel, it can be routed through Horizon Europe, the EIC or national programmes so that R&D budget carries part of the cost - and the prototype that proves the concept becomes the evidence in the proposal.
We write those proposals ourselves. It is not a referral to a partner, and it is not a bolt-on: the people who scope the architecture write the work packages.
Insights
Recent writing
Funding landscape
Reading an Impact section the way an evaluator does
Most proposals lose points in Impact, not in Excellence. The reason is structural, and it is fixable in a week.
28 July 2026 · 4 min
Applied AI
What an agentic system actually costs to run
Demos are priced per call. Production is priced per task, and the difference between the two is where most AI business cases quietly fail.
14 July 2026 · 3 min
Next step
Tell us what you are trying to decide, build, or fund.
A first conversation costs nothing and usually ends with a straight answer about whether we are the right people for it.