AI approach
Choosing the right AI, not the most AI.
The goal is not maximum autonomy. It is the simplest architecture that reliably creates the required business value.
A way of deciding where and how to use AI, drawn from my work across:
- Enterprise AI strategy and readiness
- Opportunity discovery
- Workflow redesign
- AI solution architecture
- RAG and enterprise search
- AI-powered applications
- Governance and human oversight
- Evaluation and adoption
- Productisation and scaling
Solution spectrum
More autonomy needs more control.
Eight solution patterns, from deterministic software to multi-agent orchestration. Select one to see what it is for.
Lower autonomy Lower coordination complexity
Higher autonomy Higher control and observability requirements
Which pattern fits?
Six questions about a use case. The likely pattern lights up on the spectrum.
Likely pattern
Not answered yet
Answer any question to see where a use case sits on the spectrum.
Decision path
How it decides
- No AI needed: traditional software, or workflow automation when it acts across systems or along an unpredictable path.
- AI that only informs: RAG when it needs enterprise knowledge, otherwise a single LLM task.
- AI that acts along a predictable path: an AI-powered app within one system, workflow automation across several.
- AI that acts along an unpredictable path: an agent within one system, an agentic workflow across several.
- Multi-agent orchestration only when that path also needs enterprise knowledge and the risk is lower.
- High risk adds approvals, audit trails and named accountability to whichever pattern fits.
An illustrative aid, not a recommendation engine. Real decisions also weigh data quality, cost and how the work is organised.
Transformation
From AI experiments to repeatable capability.
Choosing a pattern is one decision in a longer sequence. This is the order I work in.
-
Direction
Outcomes, priorities and boundaries.
-
Discover
Workflows, decisions and friction.
-
Prioritise
Value, feasibility, risk and reuse.
-
Design
Target workflow, UX, controls and architecture.
-
Prove
Value, quality, adoption and cost.
-
Scale
Productise, reuse and continuously improve.
Principles
Six principles for AI transformation
- Start with the business outcome.
- Discover opportunities in workflows, not AI feature lists.
- Choose the minimum powerful solution.
- Keep humans accountable where decisions matter.
- Measure value, quality and adoption.
- Scale evidence, not enthusiasm.
In practice
Where this shows up in my work.
-
Building an AI Solutions practice
Leading the development of an AI Solutions practice that connects product strategy, transformation and delivery, starting from organisational problems rather than models.
-
AI-assisted B2B product discovery
Professional buyers know the job they need done, not every product it takes. Designing discovery that starts from intent, with AI assisting expertise rather than replacing it.
-
Enterprise finance intelligence
Designing a conversational interface for complex enterprise financial data, without outsourcing calculation and trust to a language model.
-
Designing a trusted enterprise AI answer layer
Turning fragmented content, product data and enterprise systems into one governed intelligence layer for people, search and AI.