Short answer
An executive AI workshop is a structured session in which a leadership team or board decides where AI fits in the business, what risks it will accept and which one or two initiatives to fund first. Levy Street runs decision workshops for New Zealand leaders using your own documents and operating numbers, ending with a ranked shortlist, an owner for each item and a business case outline.
- The output is decisions: a ranked shortlist, owners, risk limits and next steps.
- Live demonstrations use your own documents, so leaders see real accuracy.
- Only 15% of larger NZ organisations were scaling AI organisation-wide in 2026.
- Regulators now expect boards to know where AI is used and who is accountable.
What an executive AI workshop is for
An executive AI workshop exists to get a leadership team or board to decisions: where AI fits in the business, what risks it will accept, who owns what, and which one or two initiatives to fund first. It is not a talk about the future of AI. Every exercise uses your own documents and operating numbers, and the session ends with a written decision record.
It suits leadership teams under pressure to “do something about AI”, boards setting oversight for the first time, owner-operators deciding where to spend a limited budget, and public sector leaders working within the Public Service AI Framework.
Why leadership teams need one now
Adoption has outrun leadership. Datacom’s 2026 survey of 207 senior leaders at NZ organisations with 100 or more staff found 91% using some form of AI, but only 15% scaling it organisation-wide, only 22% with a dedicated AI leadership role, and only 20% with a dedicated AI budget and clear cost tracking (Datacom). A third said returns had exceeded costs.
Oversight expectations are rising. On 15 October 2026 the Reserve Bank and FMA told regulated entities that “boards and senior management need to know where AI is being used in their business” and who is accountable (RBNZ). The National Cyber Security Centre’s Cyber Threat Report 2026 asks boards and chief executives whether they have the people and processes for an AI-accelerated threat environment. Policy may also shift: Labour campaigned in October 2026 on an AI action plan and an Office of AI (RNZ).
The decision workshop agenda
| Block | Purpose | Output |
|---|---|---|
| Where AI is already used | Replace assumptions with the pre-work inventory | Known uses, approved and not |
| Live demonstration | AI on your own contract, report or emails, including where it fails | Shared, realistic view of accuracy |
| Opportunity screen | Score candidate workflows on value, feasibility, risk and data readiness | Ranked shortlist with the arithmetic |
| Risk appetite | Decide what AI may never do, and what needs a person’s sign-off | Draft risk statements |
| Build, buy or partner | Test each shortlisted item against products you already own | A route for each item |
| Commit | Choose one or two items, owners and the number each must move | Decision record |
An example business case
Illustrative example, every figure an assumption: a 45-person Wellington professional services firm writes 30 proposals a month. The workshop tests AI drafting of proposal research and first drafts.
| Line | Assumption | Annual value |
|---|---|---|
| Time saved drafting | 2.5 hours a proposal, 360 proposals a year, NZ$85 an hour internal cost | NZ$76,500 |
| Added review time | 0.5 hours a proposal for a senior person to check the draft | minus NZ$15,300 |
| Model fees | 200,000 input and 20,000 output tokens a proposal on Claude Opus 5.5 at US$4 and US$20 per million (listed 19 October 2026); US$1.20 a proposal, US$432 a year at NZ$1.7656 per US dollar (16 October 2026) | minus NZ$763 |
| Other running costs | Hosting, monitoring and support | minus R |
| Build cost | One-off | B |
Annual net benefit is NZ$60,437 minus R, and payback in months is B divided by one twelfth of that figure. The case rests on one assumption: if real time saved is 1 hour rather than 2.5, net value falls to NZ$15,300 a year before R, and the case may not stand. The workshop’s job is to name that assumption, and the proof of concept’s job is to measure it. Prices from Anthropic; exchange rate from Frankfurter, using European Central Bank data. Our AI business case template and ROI calculator follow the same structure.
What we deliver
- Pre-read
- A short brief on what current AI can and cannot do in your sector, with dated facts.
- AI inventory summary
- Where AI is already used across the organisation, from the pre-work survey.
- Ranked shortlist
- Candidate workflows scored on value, feasibility, risk and data readiness.
- Risk appetite statements
- Draft limits on what AI may do, ready for board adoption.
- Decision record
- What was decided, by whom, with owners and the measure for each item.
- Business case outline
- For the top item, in the format above, with assumptions marked for testing.
How the work runs
- Discover: pre-work. Short interviews with a few leaders, an inventory survey, and a handful of your documents for the demonstrations. We check the products you already own, because the best first step is sometimes a feature you are already paying for.
- Prove: the workshop. Demonstrations on your material, the opportunity screen, risk appetite and commitments, run in person.
- Scale: follow-through. The decision record and business case arrive within days. The top item usually moves to an AI readiness assessment or an AI proof of concept.
Cost drivers and limits
Cost depends on the number of leaders involved, how much pre-work and interviewing is wanted, whether demonstrations are built on your own documents (they should be), and travel outside Wellington. A workshop is not a strategy: it cannot assess data quality across every team or sequence a multi-year programme, which is what AI strategy and advisory is for. If you already know the workflow you want to improve, skip the workshop and go straight to a proof of concept. Workshops also go wrong when they become a vendor showcase; ours show failures alongside successes on purpose.
Work this draws on
- Bebop: an AI sales agent for small and medium businesses, owned and sold by Blackpearl Group (NZX/ASX: BPG). Levy Street engineers work with Blackpearl Group’s AI team on Bebop, so we bring the view from inside a listed company shipping AI to paying customers.
- GTM-Bench: co-authored by Levy Street team members, it found several general-purpose AI agents scored below zero on real prospecting tasks once weak results were penalised (paper), a useful corrective to vendor demos.
Sources
- 2026 State of AI Index: AI uptake at 91%, leadership lags, Datacom, 2026-08
- RBNZ and FMA outline supervisory expectations on cyber and AI risks, Reserve Bank of New Zealand, 2026-10-15
- Cyber Threat Report 2026: leaders need to prepare now for the impact of AI, National Cyber Security Centre (NZ), 2026-09-24
- Taking a light touch with AI’s dangers and evils, RNZ, 2026-10-12
- Pricing (Claude API), Anthropic, accessed 2026-10-19
- USD to NZD reference rate, 16 October 2026 (European Central Bank data), Frankfurter, 2026-10-16
- GTM-Bench: Evaluating Agentic AI Systems for Buyer/Seller Coherence in Go-to-Market Workflows, Blackpearl Group, 2026-06
Written by the Levy Street team in Wellington. Last updated 19 October 2026.