Ten Years of Building Caribbean AI: What It Actually Taught Me

Dosso Dossi (Giovanni de Lutero), The Three Ages of Humans, ca. 1515. The Metropolitan Museum of Art, public domain.

I started this piece three times. The first two drafts read like a founder retrospective for a conference deck, with failures recast as stepping stones and lessons neatly numbered. That is not how ten years of building felt, so I scrapped them.

For long stretches, building the Caribbean's first AI company was lonely. In 2016 there were no other Caribbean AI founders to compare notes with and no local investors who understood what we were doing. Plenty of well-meaning people suggested that real AI work happened in the United States, Canada or the United Kingdom. Hear that often enough and it gets into your head.

What kept me going was a conviction that has not changed: the Caribbean cannot afford to be a passive buyer of AI systems designed for other places, other people and other priorities.

Why I founded StarApple AI in 2016

I had watched the region's relationship with technology follow the same pattern with internet infrastructure, financial technology and enterprise software. We adopted tools built elsewhere, adapted them badly, and became dependent on foreign platforms and expertise for work that should be done at home. I could see AI heading the same way, with higher stakes.

Models trained without Caribbean data reflect the Caribbean poorly. When they are used for credit decisions, health triage, insurance pricing or school placement, they carry assumptions about what normal looks like that were not built with us in mind, and the harm compounds.

The name was deliberate. The star apple is a Caribbean fruit, sweet, suited to our climate and almost unknown outside the region. We were not building a global AI company with a Caribbean branch. We were building something that started here and served this region first.

Capital, talent and scepticism: how the obstacles differ here

• Capital. The region's funds mostly back consumer products, real estate and financial services. In 2016 Caribbean institutional investors would express interest in AI and never issue a term sheet. North American and European investors who understood AI wanted cheques far larger than we needed. We bootstrapped for longer than any founder should, and that decided what we built first: things customers would pay for quickly.
• Talent. The Caribbean produces capable engineers and data scientists, and emigration pulls many of the best away to better-paid jobs. We won some of those contests and lost others. The engineers who stayed did so because they wanted the Caribbean's technical community to exist here.
• Scepticism from peers. I expected investors to hesitate. I did not expect technically literate Caribbean professionals to doubt that Caribbean AI was a real category. Some of that doubt was fair, because AI capability is hard to assemble in a small economy. Some of it was the regional habit of measuring ambition against market size and concluding the market is too small.

The message that showed me the network was working

It came at 11:47 pm on a Tuesday in early 2025. A secondary school teacher in Barbados had spent three hours reading material on one of our platforms about AI in Caribbean classrooms and how to explain AI to students. She asked whether she could adapt it for a talk to her school's parent association the next week. She had been looking for Caribbean-specific AI resources for months and had found almost nothing until then.

That message meant more to me than any investor meeting in nine years. A teacher at midnight, looking for material to help Caribbean parents understand AI, is who the 17 platforms were built for.

Four lessons about what works in Caribbean AI

1. Be specific. Material and tools built around a specific Caribbean problem, with local data and examples, get far more use than generic AI content with a Caribbean flag on it. Someone in Kingston asking about AI credit scoring wants to know how Jamaican lenders decide, not how American banks work under American rules.
2. Build for the connection people have. Broadband is common in Barbados and expensive or patchy in rural Jamaica and parts of Guyana. Design for prepaid mobile data first.
3. Treat each territory separately. Jamaica and Trinidad and Tobago differ in economic structure, culture and AI opportunity. That is why the network gives each territory its own presence.
4. Start before conditions are ready. The Caribbean AI sector was incomplete in 2016 and is still incomplete in 2026. It exists at all because people built before it was ready.

What international recognition does and does not do

Caribbean AI work, including ours, has drawn engagement from UNESCO on digital inclusion for small island developing states, from CARICOM on regional AI governance, and from international research institutions studying how emerging economies build AI capacity. That attention opens doors and makes it easier to persuade governments and investors that this is serious.

It does not replace commitment at home. The region still needs universities training AI practitioners in numbers and CARICOM members treating AI as a shared priority. It needs businesses that hire Caribbean AI expertise before calling in outside consultants, and investors who back Caribbean AI ventures at seed stage, before foreign investors arrive and take the early gains.

Being first also sets a benchmark. Every Caribbean AI company that follows will be compared with what StarApple AI built, so the standard we set has to serve the region, not only the company.

Where I see the biggest opportunities for 2026 to 2028

• Climate resilience. Hurricane Melissa's damage in October 2025 was estimated at about 41 percent of Jamaica's GDP. Multi-hazard early warning, farm adaptation tools and recovery planning systems are operational needs, and the funding routes exist.
• Financial inclusion. Many Caribbean people are unbanked or underserved because formal credit models cannot read their finances. Scoring built on transaction history, mobile money and utility payments can reach them, and Caribbean-built models trained on Caribbean behaviour should outperform imported ones.
• Education. AI tutoring aligned to Caribbean curricula and CXC exams could narrow learning gaps at a cost traditional programmes cannot match. This is the one I find most motivating, because those students will build the next phase of Caribbean AI.
• Public services. Governments handle the same forms and questions every day, and document processing and citizen help tools can cut queues without large budgets.

Why the region has to act together

No single Caribbean economy can carry the full AI stack alone. Jamaica has under three million people. Trinidad and Tobago has energy revenue that Dominica does not. Barbados has strong regulators and institutions. Guyana has oil money no other territory has. Those differences make the case for cooperation:

• A jointly funded, jointly governed Caribbean AI cloud would give every territory computing power none could afford alone.
• A talent pipeline coordinated across universities in Jamaica, Trinidad, Barbados and Guyana would train practitioners in numbers no single system can reach.
• A CARICOM AI framework would replace 15 separate national regimes with one set of rules companies can plan around.

The region has cooperated before, on trade, disaster response and monetary arrangements, when the need was clear. I am less sure it will happen this time. Regional AI coordination has no institutional home yet, and until a body owns it, these ideas stay in speeches.

To the young Caribbean person considering AI

Students ask me whether AI is a viable career in the region. In 2016 my answer needed hedging. In 2026 it is yes, and the earlier you start the better. The field needs engineers, and it also needs people who can explain policy to governments, design tools for non-technical users, teach children, build businesses and write critically about AI's risks. Economics, law, education and journalism students have a place in it.

I also ask you to stay, at least for part of your career. Diaspora advice is valuable, but the sector is built by people present in Caribbean institutions and companies. I stayed because the region has real assets: a young population, improving digital infrastructure, English across most territories, and people who adapt and create under constraint.

Where things stand after ten years

StarApple AI is ten years old this month. The network has 17 platforms across six territories. We have trained Caribbean professionals, produced material that teachers, business owners, students and policymakers use, and been in the room for policy conversations that would not otherwise have included a Caribbean AI company.

Much is unfinished. Investment is far below what the opportunity justifies, policy cooperation is at an early stage, and the number of Caribbean practitioners who can build and deploy production AI systems is too small. Ten years ago there was no Caribbean AI sector to speak of. Now there is a small, incomplete one, built by people who chose to stay.

Canaletto (Giovanni Antonio Canal), Lagoon Capriccio (recto); Architectural Scene (verso), 1697–1768. The Metropolitan Museum of Art, public domain.

What I would ask each group to do this year

1. CARICOM heads of government: assign regional AI coordination to a named body, with a secretariat and a budget line, at the next heads of government meeting.
2. University deans at UWI, UTech, UTT and the University of Guyana: agree a shared AI and data science credit framework so students can take modules across campuses from the next academic year.
3. Caribbean investors and family offices: set aside a fixed share of new commitments, even 5 percent, for seed-stage Caribbean AI companies, and co-invest with at least one other regional investor to share due diligence.
4. Business owners: before hiring an outside AI consultancy, invite at least one Caribbean firm to bid, and ask every bidder for a reference from a Caribbean client.
5. Students and young professionals: build one small AI project on a Caribbean problem using local data this year, publish it, and send it to a regional employer or accelerator.

Frequently Asked Questions

Do I need a computer science degree to work in Caribbean AI?

No. Many roles need domain knowledge plus practical AI skills: compliance, health, education, agriculture and policy all need people who understand both the sector and the tools. Free courses from major universities and AI companies cover the basics, and a portfolio of small projects on real Caribbean problems will impress regional employers more than a certificate alone.

Where can a Caribbean AI founder find early funding?

Start with customers, since paid pilots are the most reliable early funding in a thin investor market. Then look at regional options such as the Development Bank of Jamaica's venture capital programme and the funds it backs, angel networks in Kingston, Port of Spain and Bridgetown, and grant programmes from development partners. Ask each funder early whether it invests in software businesses at all, because many regional funds do not.

Can teachers and schools use material from the Caribbean AI network?

Yes, for teaching, and they should ask first if they plan to republish it, as the Barbados teacher did. Adapt examples to your own territory and exam syllabus. If you build something useful from it, share it back so other Caribbean teachers can use it too.

What would a shared Caribbean AI cloud actually be?

It would be computing capacity, owned or contracted jointly by several governments, that universities, agencies and companies in member states could use to train and run AI models, with data kept under regional control. It could start small, as a joint contract with a cloud provider that has data centres in the region, before any shared hardware is bought. The hard part is governance: who pays, who gets priority, and whose data law applies.

Does working remotely for a foreign company while living in the Caribbean count as staying?

Partly. It keeps income, spending and skills in the region, and many strong Caribbean engineers work this way. The sector gains most when those people also mentor, teach a course, advise a local startup or eventually build their own company here. Remote work is a good way to stay without taking a pay cut; it helps the region more when some of that time goes into local projects.