AI strategy
Why AI pilots stall between demo and production
The gap is rarely model capability. It is unit cost, data readiness, evaluation, and an unanswered question about accountability.
7 min read
Practice 03
AI Strategy
Decide where AI actually earns its cost, before you spend the year finding out.
Engagement shapes
Strategy sprint
3 weeks
Full strategy
6 to 8 weeks
Executive workshop
2 days
Advisory retainer
Monthly
Who this is for
- CIOs and CTOs holding a budget and a portfolio of disconnected experiments
- Chief Data and AI Officers asked for a position by the board
- Executive teams choosing between building a platform and buying one
The problem
The constraint is rarely model capability. It is choosing which problems are worth automating at all.
The questions that decide the outcome are unglamorous: what data is actually usable, what a unit of inference costs at production volume, and who is accountable when the output is wrong.
A strategy that answers none of those is a use-case longlist with a cover page.
What we do
The workstreams in an engagement
01
Opportunity mapping
Across the value chain, function by function, against where cost and cycle time actually sit.
02
Portfolio scoring
Use cases scored on value, feasibility, data readiness, and risk, then sequenced rather than ranked.
03
Data readiness
Assessed for the shortlisted use cases only, so the work stays finite.
04
Build, buy, or partner
Decided per capability, with the reversal cost of each named.
05
Reference architecture
Model-agnostic by default, so a provider change is a configuration change rather than a rewrite.
06
Unit economics
Cost per task modelled at pilot volume and at production volume, handed over as a model you can re-run.
07
Governance and policy
Mapped to NIST AI RMF, ISO/IEC 42001, and the EU AI Act where it applies to your products and markets.
08
Investment case
Staged funding with gates that can genuinely stop a workstream.
Deliverables
What lands on your desk
You keep all of it, including the method behind it, so the work can be repeated without us.
- AI strategy document stating decisions, not options
- Scored and sequenced use-case portfolio
- Reference architecture and platform recommendation with the trade-offs stated
- Governance framework, risk taxonomy, and acceptable-use policy
- Unit-economics model you keep and can re-run
- Investment case with stage gates
Standards we work against
- NIST AI Risk Management Framework
- ISO/IEC 42001
- EU AI Act
- ISO/IEC 23894
What we do not do
- We do not produce a use-case longlist and leave.
- We do not recommend a model vendor we cannot justify on cost and evaluation evidence.
Engagement shapes
Three sizes, not one package
Durations are indicative and depend on estate size. Scope and price are fixed in writing before the engagement starts.
| Shape | Duration | What you get |
|---|---|---|
| Strategy sprint | 3 weeks | Shortlist, economics, and a direction to commit to |
| Full strategy | 6 to 8 weeks | Portfolio, architecture, governance, and investment case |
| Executive workshop | 2 days | Shared framing and a decision log |
| Advisory retainer | Monthly | Standing counsel as the portfolio moves |
Related insights
Our thinking on this
Next step
Start with a briefing, not a proposal.
Thirty minutes on ai strategy. We will tell you whether we are the right firm for the problem, and who to talk to if we are not.