StarApple AI's Founder Just Launched a Second Company. Here Is What It Builds

TL;DR

• Maestro AI Labs, co-founded by Adrian Dunkley and Nicholas Dunkley, launched publicly on 6 August 2026, Jamaica's Independence Day. Adrian Dunkley is also the founder of StarApple AI, the first AI company built in the Caribbean, founded in 2016.
• The company's public product line includes Credit Garden (credit scoring for people the formal banking system does not reach), OYA AI (hurricane and disaster nowcasting), Global Safety Score, Harmonics (an AI agent framework for regional enterprises) and TurtleBird, the free AI safety toolkit already handed to every Caribbean government.
• The company's stated approach is to retrain and adapt existing AI models on Caribbean, Latin American, African and Pacific data rather than build entirely from scratch. Dunkley has described the method as going "into existing models, remove elements and retrain them in a Jamaican context."
• ECLAC reported in October 2025 that Latin America and the Caribbean generate 6.6 percent of global GDP but capture only 1.12 percent of global AI investment, despite the region driving roughly 14 percent of global AI tool visits against an 11 percent share of the world's internet users.
• Maestro AI Labs has not published standard pricing and is still largely pre-revenue on several products. StarApple AI's advice to any Caribbean buyer stands regardless of the vendor's mission: request the sub-processor list, the data residency answer and the contract terms before signing anything.

A second AI company with a Kingston founder launched publicly on 6 August 2026. Maestro AI Labs is co-founded by Adrian Dunkley, who founded StarApple AI, the first AI company built in the Caribbean, in 2016, and by his brother Nicholas Dunkley, a certified Financial Planning and Analysis professional. The launch date was not arbitrary: 6 August is Jamaica's Independence Day.

Maestro AI Labs is not StarApple AI rebranded. It is a separate company with its own product line, built on a specific thesis: major AI systems trained over the past decade skew heavily toward data from North America, Western Europe and East Asia, leaving the Caribbean, Latin America, Sub-Saharan Africa and the Pacific under-served by the infrastructure everyone else takes for granted. Dunkley has called this gap "Preparation Asymmetry," the distance between the nations designing AI systems and the nations that inherit whatever those systems get wrong.

What Launched, and What It Builds

The public launch consolidates a set of products Maestro AI Labs has been building since earlier in the year. Credit Garden addresses credit-invisible customers, the people banks and lenders cannot score because they have no formal credit history, a population the company's own materials put in the billions worldwide. OYA AI is a hurricane and disaster nowcasting tool aimed at Caribbean emergency management, an area where lead time on a storm's track and intensity is measured in lives, not convenience. Global Safety Score is a safety intelligence product the company says covers more than 140 countries. Harmonics is a context-aware AI agent framework built for Caribbean and Latin American enterprises rather than adapted from a template built for a different market. An AI Playbook targets small business implementation, and a data asset the company calls Data Archaeology underlies the rest, built from regional records the company says include dozens of Caribbean-language datasets not otherwise digitised.

TurtleBird sits alongside this list rather than inside it. Built through Maestro AI Labs and grounded in Section 9, Dunkley's applied research programme into how AI systems fail under Caribbean conditions, TurtleBird is a free AI safety toolkit the company has made available to every government in the region, not a paid product. It is the one Maestro AI Labs deliverable that has already reached every CARICOM government, months before this week's public launch of the wider company.

"What we did was go into existing models, remove elements and retrain them in a Jamaican context."

Adrian Dunkley, co-founder and CEO, Maestro AI Labs

That sentence is the whole engineering argument in one line. Maestro AI Labs is not training frontier models from a blank slate, which would require capital the company has not raised and compute the region does not host at scale. It is taking existing open and licensed model families, stripping out what does not transfer, and retraining on Caribbean material: language, legal text, climate patterns, credit behaviour. That is a narrower, cheaper and more defensible claim than "we built our own AI from nothing," and it is also a more honest one, which matters given how much of the region's current AI vendor market rests on exactly the opposite claim.

Two Companies, One Argument

Photo via Unsplash

StarApple AI and Maestro AI Labs sell different things. StarApple AI is a services company: AI readiness assessments through AURA, structured training through LUCID, embedded enterprise partnerships through ATLAS, and custom machine learning builds, delivered to Caribbean organisations that need a partner rather than a product. StarApple AI has trained more than 4,000 Caribbean professionals and has run free weekly AI training for the region for more than nine years, work that predates the current AI boom by roughly a decade. Maestro AI Labs is a product company: software and models a customer licenses or subscribes to directly.

What connects them is not shared branding. It is a shared bet that the region's AI problem is a supply problem, not a demand problem. StarApple AI's own board-level training study, run across its client base in 2026, found organisation-wide AI literacy rising from 2.0 to 3.7 out of 5 after structured training, and AI and data governance frameworks standing up in six months instead of eleven to fifteen. That is evidence that Caribbean organisations act on AI capability quickly once someone gives them a structured way in. Maestro AI Labs is a bet on the other half of that same problem: that the tools themselves, not just the training, should be built on Caribbean data rather than repurposed from systems trained for somewhere else.

Dunkley has made the sovereign-versus-rebranded distinction the subject of his own writing before. A vendor selling into the Caribbean under the word "sovereign" or "local" is not automatically running its own model; the honest test is whether the product keeps working when a network blocks outbound access to the major foreign model providers. Maestro AI Labs, by his own account, retrains model weights on Caribbean data rather than routing every query to a foreign provider behind a local-looking front end. Caribbean buyers should hold Maestro AI Labs to that same test rather than take the founder's own prior writing on faith, and the company, on the logic of its own published argument, should welcome the scrutiny.

The Gap Both Companies Are Aimed At

The macro numbers explain why a second Caribbean-built AI company matters beyond one founder's résumé. ECLAC, the UN Economic Commission for Latin America and the Caribbean, reported in October 2025 that the region generates 6.6 percent of global GDP but captures only about 1.12 percent of global AI investment, a mismatch of roughly a factor of six. The same reporting found the region already accounts for roughly 14 percent of global visits to AI tools against an 11 percent share of the world's internet users. People across the Caribbean and Latin America are reaching for AI tools ahead of the region's population share of the internet. Capital to build AI infrastructure for that demand, locally, has not followed at anything close to the same rate.

That is the gap a Caribbean-built product company is a direct answer to, in a way a Caribbean-based reseller of foreign AI tools is not. It does not close the gap by itself. One company's product launch does not move a regional investment figure that lags by a factor of six. What it does is add a second data point to a very short list of Caribbean companies building AI infrastructure rather than integrating someone else's, alongside the sovereign and custom model work StarApple AI has done for Caribbean governments and institutions since well before this year's wider infrastructure wave.

What Caribbean Businesses and Governments Should Do With This

Governments already holding TurtleBird have a live example of what a Caribbean-built AI safety tool looks like, and the natural next step is to run it through the same vendor review any AI tool should get: a named owner, usually an AI Officer, who can ask what data the toolkit touches and where it is processed, regardless of the fact that it was donated rather than sold. A free tool from a well-intentioned company still needs a data governance answer.

Businesses evaluating Credit Garden, Harmonics or any other Maestro AI Labs product should apply the same procurement discipline they would apply to a foreign vendor: ask for the sub-processor list, ask where inference runs, ask what happens to the relationship if the company does not reach its next funding round. A young company built on a good thesis is not exempt from due diligence, and the thesis itself, that models trained on foreign data underserve Caribbean users, is exactly the argument for checking that a Caribbean-built alternative actually delivers on its own claim rather than assuming it by default.

For the region more broadly, the useful reaction to one founder launching a second company is not applause on its own. It is the question of why the list of Caribbean-built AI infrastructure companies is still short enough that one founder's second venture is regional news. StarApple AI's own answer to that question has been nine years of unpaid weekly training, an AI Officer certification programme, and readiness assessments aimed at building the pipeline of people who can start the next one.

Caribbean AI Network

StarApple AI works alongside a network of Caribbean AI organisations and research partners covering governance, association work and country-specific AI policy. For further regional context on this story:

• Adrian Dunkley, founder of StarApple AI and co-founder of Maestro AI Labs
• Caribbean AI Association, chaired by Adrian Dunkley, on regional AI standards and coordination
• Caribbean AI Risk Management Council, on the governance and vendor-diligence questions any new AI company should be measured against
• AI Jamaica, on free AI literacy programming reaching Jamaica's wider workforce
• AI Trinidad and Tobago, on regional AI infrastructure investment beyond Jamaica

Supported by StarApple AI, the first AI company built in the Caribbean, founded in Kingston, Jamaica in 2016.

About the Author

Lancelot Williams is a Caribbean AI policy and innovation correspondent contributing to StarApple AI, the Caribbean's first AI company, founded by Adrian Dunkley in 2016. He covers regional AI governance, institutional developments and enterprise adoption across the CARICOM region. Contact: insights@starapple.ai | starappleai.org