Consulting innovation portfolio

Making quality legible,
before AI made it urgent

The AI work on this site has a twenty-year ancestry. Long before eval scores and drift metrics, the same problem existed in every testing engagement: the people who owned the outcome could not read the evidence. These nine innovations are how that gap has been closed, from the consulting years into the AI era. Each page shows why the technique exists, how it works, what it changed, and the AI-era build it connects to.

Quality Integration Maps

Customer, business, and technology priorities on one page, mapped against the customer journey. Test priority set by what matters, visible to everyone at once.

Read →

Sankey scope review

Priorities flow into questions, questions into testing activities, thickness carries importance. A $30M+ programme GM called it the clearest view of testing she had ever seen.

Read →

Technical due diligence in procurement

Between shortlist and pilot: cover the biggest risks, compare suppliers blind, and qualify the field down before the six-month commitment.

Read →

Behaviour trees

End-to-end behaviour in one formal model. Two SMEs discovered they had signed off on different readings of the same words, minutes into a playback.

Read →

One-page executive reporting

The exec-coffee-cup test. Verdict-first reports, three-page plan summaries, and activity handouts that replace thirty pages with four.

Read →

Three-lens risk workshop

Pre-mortems across customer, business, and technology lenses, producing the questions testing must answer, in words the room actually owns.

Read →

Improvement opportunities map

Recommendations on one page, grouped by owner, with dependency rings showing which items unlock the rest. The review artefact that survives the review.

Read →

Global quality operating model

Delivery, governance, tooling, sales, capability, and people as one system, built for a 650-person practice across four countries, and reused beyond it.

Read →

Dual-track AI assurance

Application testing proves the system, AI assurance proves the intelligence, and both converge to one standard. A breach after go-live re-enters the loop.

Read →

Coverage gap audit

Which quality attributes matter for a system's DNA, which are actually measured, and which gaps a regulator would ask about. Deterministic on purpose: every finding is a checkable lookup, not a score.

Read →

Clarification questions

For a system not yet ready to be scored, a verdict is the wrong artefact. Turn the gap audit into the questions a team must answer first, ranked by the impact of getting each one wrong.

Read →