For the AI decision before any team gets hired
AI Consulting
Every AI budget eventually meets a harder question than the technology itself: which use case actually deserves the investment. This practice answers that question with evidence instead of enthusiasm, whether the pressure came from a board slide or a competitor's announcement. Ranked cases, tested feasibility and a costed recommendation replace guesswork before any build gets funded.
How an AI recommendation actually gets made
Ranking, feasibility testing and the costed case that reaches your board.
90 secSample frame · video pending01Agreed before any workshop
The ledger that keeps the assessment honest
Every consulting engagement fixes the split before work starts: what you bring, what we assess, what comes back, and what never leaves your control regardless of which case wins.
- Your side
You bring
- The pressure and any early AI ideas floated
- Access to the teams closest to candidate workflows
- A rough sense of budget appetite for a first case
- Our side
We assess
- A structured review of every candidate use case
- Feasibility testing against your data quality
- A value and effort score attached to each candidate
- Your side
You receive
- A ranked shortlist with a costed first case
- A written feasibility read for every candidate
- A recommendation your board can approve or decline
Never leaves you
- Which case, if any, gets funded first
- All commercial and roadmap decisions that follow
- The right to stop after the shortlist
Where the line sitsWe rank and cost the options. Whether to fund one, and which, stays a decision only you make.
02Evidence and proof
Recommendations a board actually funded, not just admired.
Consulting proof is a decision that got funded and a case that survived contact with real data, not workshop attendance. Every record here waits for written client approval before it replaces this preview.
Explore case studiesA ranked shortlist turned a stalled AI mandate into one funded case.
Aster North had six competing AI ideas and no way to choose. A structured ranking and feasibility read against their own data picked the case worth funding first.
- Candidate cases assessed
- Six, scored and ranked
- Cases recommended for funding
- One, with a costed plan
A build-versus-buy call saved a learning platform months of engineering.
Pillar Learning was scoping a custom recommendation engine. The comparison showed a configured vendor tool met the need at a fraction of the build cost.
- Build effort avoided
- A full engineering quarter
- Time to a working system
- Weeks, not months
They told us which of our four AI ideas to kill first. That was worth more than being told which one to build.
Sample · pending client approval
03Read this first
When consulting is the wrong starting point
Selling a ranked case list to a team that already knows what to build would cost more than it earns us in trust. Two alternative starting points are marked before any pitch.
- AI Software Development
You already have a costed, proven case
Skip the ranking exercise and go straight to engineering it into production software.
- AI Transformation
The whole organisation needs to change, not one case
A single ranked recommendation cannot carry an organisation-wide shift. That work runs as a transformation programme instead.
- Hire AI Engineers
You need engineers, not another decision
If the decision is already made and the gap is capacity, hiring AI engineers gets you moving faster.
| Shape | Best when | You direct | We carry |
|---|---|---|---|
| Scoping conversation | The pressure is real but shapeless | Which workflows to discuss | An initial view worth ranking |
| Ranking workshop | Several candidates compete for funding | Priorities and constraints | Scoring, feasibility, the shortlist |
| Full assessment | A major decision needs board sign-off | Final approval | Ranking, testing, cost modelling, recommendation |
04Straight answers
Asked before AI budgets get signed.
01How is AI consulting different from AI software development?
Consulting decides what deserves to exist: ranked cases, feasibility against your data, and a costed recommendation. Software development engineers the chosen case into production. They run as one practice, so nothing gets lost in the handoff, and the same team stays accountable for what the recommendation produces.
02Do you recommend vendors you have a relationship with?
No. Vendor and model recommendations are scored against your constraints, not a partner list. Where a vendor is the right call, that gets said plainly, even when the alternative would have been more work for this practice to deliver.
03What if the ranked case list says nothing is worth building?
That is a valid, useful outcome, not a failed engagement. Some years the right AI decision is to wait, and a report that says so honestly saves a build that would have failed later at far greater cost.
04How long does a consulting engagement take?
A scoping conversation runs in days. A full ranking and feasibility assessment across several candidate cases typically closes within a few weeks, timed to when your board or budget cycle actually needs the recommendation.
05Can you assess a case a competitor already sold us?
Yes, and it is a common request. A second, independent read on a vendor's proposed AI case is often the fastest way to confirm the recommendation or catch what the first pitch left out.
06Do you test feasibility against our actual data?
Always, on a sample basis, before any recommendation is written. A case that looks strong in theory can fail once real data quality, access rules and edge cases are accounted for, so the test happens early.
07What happens after the recommendation is delivered?
You decide what happens next. Some clients take the recommendation to their own team, some commission the build here, and some pause deliberately. All three are honoured outcomes of the same engagement.