Board-Level AI Training Fuels Company Success, a Caribbean Study Finds

When a board learns how AI actually works, the organisation it governs performs differently, and a 2026 study now puts numbers on the difference. StarApple AI and Partners, the Jamaican firm behind more than 100 boardroom training engagements across the Caribbean, went back to the boards it trained and scored them. Those organisations lifted AI literacy from 2.0 to 3.7 out of 5, doubled deployed AI initiatives in eight months, and cut vendor costs by over 70 percent.

The study is unusual for this region. Caribbean AI coverage tends to run on projections borrowed from other markets. This one draws on organisations headquartered here, scored before and after their directors sat through board-level AI training, across an eight-month observation window on internal indices out of 5.

Every measure below moves the same way, toward boards that knew enough to act on judgement instead of nerves. The change began at the board, not in the IT department, and worked its way down from there.

01Literacy moved first, and it moved from the top down

The study's central measure is an AI literacy index scored out of 5. Across the trained organisations, the organisation-wide score rose from 2.0 to 3.7 over the study period. The researchers attribute the rise to a sequence: once boards understood the technology, they approved programmes, budgets and experimentation that pushed capability down through business lines to people managers and their teams.

The sharpest improvement was with the directors themselves. Board data literacy rose from 1.8 out of 5 to 4. The firm links that jump to the collapse of an old barrier: directors no longer needed to write code to work with data. Trained board members ran their own analysis, built working prototypes with AI coding tools, and translated information across functions without waiting for a technical intermediary.

That distinction, between an organisation's average literacy and its board's own literacy, is easy to disregard. Corporate training budgets in the Caribbean typically point downward: frontline staff learn the new system, middle managers get a briefing, and the board receives a slide deck once a year. This study measured what happens when the sequence runs the other way. The 1.8-to-4 jump at board level came first; the 2.0-to-3.7 jump across the wider organisation followed it.

“

The board is the ceiling on an organisation's AI ambition. Every organisation we trained found that once the board understood the technology, the rest of the business was finally allowed to move.

Adrian Dunkley, who led the study

02Governance stopped being an argument

Organisations took 11 to 15 months to stand up AI governance and data governance. After training, the same work took 6 months. StarApple AI credits board buy-in: directors who understood why data governance had to come first stopped treating it as an IT expense and started treating it as a precondition for everything else on the AI agenda.

The firm reports lower overall risk across the trained organisations, and says gender-related bias and equity review were built into the training and into how boards then examined AI work. This is not only an ethics point. A governance framework built without an equity lens gets rebuilt once regulators or auditors ask the questions the lens would have caught, and it was that rebuild the six-month figure avoided.

03Fewer pilots died, and the ones that lived paid

The common failure in AI adoption, pilots that demo well and never reach production, grew rarer after training. Deployed initiatives rose from two to four over eight months, more than 50 percent on the study's own framing. Time to value fell from around a year to around a month.

The study puts this down to discipline rather than enthusiasm. Trained boards understood the requirements, needs and risks of AI work, so executives and managers stopped taking on more than they could deliver and pointed their attention at initiatives with measurable returns. Communication improved in both directions, with teams using AI tools to translate and share information across functions. Several boards went further and built their own tools in-house on an agents-based approach, which StarApple AI credits with tighter board cohesion.

Two pilots reaching deployment sounds modest against four. Set against the industry baseline it is not: most AI pilots anywhere in the world never leave the pilot stage. Doubling the number that reach production, in eight months, on a base this small, is a result a board can report at the next annual general meeting.

04The vendor bill was the loudest number

Organisations in the study saved over 70 percent on vendor costs after training, savings the firm puts at tens of millions of US dollars in total. The mechanism is blunt: leaders who could not previously judge an AI pitch bought what they were sold. Once training demystified how AI systems are actually built, the same leaders could separate what their organisation needed from what a vendor wanted to sell, and priced accordingly.

“

Boards were paying for AI they did not need because they could not question what they were being sold. Once AI was demystified, vendor spend dropped by over 70 percent.

Adrian Dunkley, who led the study

For a region of small economies, the finding lands hard. Caribbean firms have little capital to sink into shelfware, and money clawed back from a padded contract is money freed for talent or deployment. Vendors with real products should want literate buyers, since the study suggests trained boards keep paying for what works.

The vendor numbers carry a second effect, less obvious than the savings. A board that can interrogate a pitch can also defend a good vendor against internal sceptics. Literacy cuts weak spending and protects the contracts worth keeping, because the people approving a renewal understand what they are renewing.

The study data at a glance

What boards should take from this