How I Work
A practical AI delivery method that reduces risk before it scales
My approach is deliberately commercial: understand the operation, design the business case, ship the smallest useful working system, then improve from real use.
Phase 01
Discover
Understand the real operating environment before recommending anything: process, people, systems, data, workarounds, ownership, and the commercial reason for change.
Decision Gate
Is there a problem worth solving, and is the data/process ready enough to support automation?
What I need from you
- Access to the people who know the process
- Example reports, spreadsheets, exports, or system screenshots
- A clear view of what is painful, risky, slow, or expensive today
Deliverables
- Current-state systems, data, and process map
- Stakeholder interview findings and pain-point ranking
- Data readiness, access, and quality assessment
- Quick-win and no-go findings before budget is wasted
Risk Controls
- No build starts until the problem and owner are clear
- Assumptions are separated from verified facts
- Simpler automation wins are considered before AI
De-risking Delivery
The point is confidence before commitment
AI projects fail when they jump from enthusiasm to build without enough commercial discipline. This approach creates decision points before spend, keeps the first delivery focused, and makes control part of the work.
Small enough to prove
The first delivery is scoped around a useful workflow, not a giant transformation programme. That makes value easier to test and easier to stop if the case is not there.
Controls are visible
Finance ownership, human review, exception handling, audit needs, and fallback routes are designed into the project rather than bolted on later.
Built around operations
The work connects to existing systems and teams. The aim is adoption inside the business, not a clever tool that lives outside the real process.
Working Principles