Part of AI consulting

Getting AI into everyday use, past the pilot

Working out where generative AI helps, choosing platforms and vendors, and getting AI from a pilot into everyday use in a way people can trust.

In my experience AI work stalls less on the technology than on the basics around it: nobody has decided what good looks like, the content isn't ready, or the people who have to support it weren't part of the decision. Strategy work is mostly about getting those right early.

Questions I help answer

A leadership team asks

“Everyone wants to know our AI plan. Where should we use it?”

What I check
Where AI would help your services most, and what a wrong answer would cost in each place.
What you get
A ranked shortlist of use cases and a roadmap leadership can act on.

A project lead asks

“Our pilot went well, then stalled. Why?”

What I check
The basics around it: what good looks like, whether the content is ready, and who will support it.
What you get
A plan to get from pilot to everyday use.

A privacy officer asks

“How do we do this responsibly?”

What I check
Privacy, accessibility and public trust, against the Public Service AI Framework and Algorithm Charter.
What you get
Those built in from the start, not checked at the end.

What it covers

  • Finding and prioritising the use cases where AI helps most
  • Vendor-neutral evaluation of platforms and vendors
  • Roadmaps and long-term recommendations leadership teams can act on
  • Responsible adoption: privacy, accessibility and the frameworks that apply, such as New Zealand's Public Service AI Framework and Algorithm Charter
Use-case shortlist Workshop 2 output Example
UseValueRiskContent readyOwnerDecision
Draft replies to common emailsContact centre, staff check every reply Service delivery Pilot now
Find outdated and duplicate pagesWeb team, before the migration Digital Pilot now
AI answers in site searchPublic, cites the official page Digital and policy After content fixes
Summarise submissionsPolicy team, internal only Policy Later
Chatbot deciding eligibilityPublic, gives a yes or no None Not now

Scoring1 to 5, with the service owners. Risk covers privacy, harm from a wrong answer, and public trust.

A use-case shortlist from a strategy workshop. An example, not one client's.

Where I've done it

  • Briefed MBIE leadership and channel teams on the direction for AI-assisted search
  • Facilitated a cross-team findability strategy workshop that helped shape the roadmap and an action plan
  • Put together MBIE's findability roadmap, which includes search, and long-term recommendations for AI-assisted search across MBIE's websites
  • Connected a network of government search and findability practitioners
  • Coordinated WCAG 2.2 accessibility audits across Health New Zealand's public and staff platforms

Findability and AI search at MBIE

Need this for your agency? I help with it as AI project consulting: where AI is worth using, choosing a platform, and getting from pilot to everyday use.

Common questions

Why do AI pilots stall before they reach everyday use?

Less often because of the technology than the basics around it: nobody has decided what good looks like, the content isn't ready, or the people who have to support it weren't part of the decision.

How do you choose which AI use cases to start with?

Rank them by how much they help the people using the service and how much a wrong answer would cost, and back the choice with test questions and evidence.

What makes AI adoption responsible?

Being clear about what the AI may and may not do, testing it before and after launch, and meeting the expectations that apply to the organisation, such as privacy and accessibility. In New Zealand government that includes the Public Service AI Framework and the Algorithm Charter.