TL;DR
- IDB Lab's May 2026 "Startups x AI" report found 85% of Latin American and Caribbean startups already use generative AI and 75% use predictive AI, concentrated in marketing, product development, and decision-making.
- A 2026 StarApple AI study found only 13% of Caribbean adults aged 18 to 65 use generative AI at all, against a global adult adoption rate of roughly 55%.
- Digicel's AI assistant Ruby has handled more than 535,000 conversations since its Jamaica launch, resolving 55% without a human agent, and is now expanding across more Caribbean markets.
- CARICOM's COTED-ICT formally endorsed the UNESCO Caribbean AI Policy Roadmap on 7 July 2026, the first collective AI policy endorsement by a CARICOM principal organ, but the document has not caught up to the deepfake fraud pattern already running in Trinidad.
- The gap between what Caribbean businesses are doing with AI and what Caribbean people are doing with it is the market 14West's founders are best placed to close.
Caribbean businesses are adopting AI faster than Caribbean consumers, and the gap between the two is wide enough to reshape where founders should build. In 2026, IDB Lab put generative AI adoption among Latin American and Caribbean startups at 85%, while a StarApple AI study found only 13% of Caribbean adults use generative AI at all, a gap the region's new policy roadmap has not closed and, in Trinidad, has not even fully named yet.
Two numbers from the same region, in the same year, rarely sit this far apart. Digicel's AI customer service agent resolved its way through more than half a million conversations in Jamaica before most Jamaicans had knowingly opened a generative AI tool of their own. That split is not a footnote. It is the shape of the Caribbean AI market right now, and it tells founders exactly where the paying customers already are.
The Adoption Split: What the Numbers Actually Say
Start with the business side. IDB Lab's "Startups x AI" report, published in May 2026, surveyed non-AI-native startups across Latin America and the Caribbean and found generative AI adoption at 85% and predictive AI adoption at 75%, with the heaviest use in marketing, product development, and day-to-day decision-making. These are companies that were not built as AI businesses folding AI into how they already operate, because the return showed up fast enough to justify it.
Now the consumer side. A 2026 StarApple AI study, led by founder and CEO Adrian Dunkley, found that just 13% of Caribbean adults aged 18 to 65 currently use generative AI tools, compared with roughly 55% of adults worldwide. The same study put the region's share of global AI investment at about 1.12%, against a Caribbean share of global GDP closer to 6.6%, and cited the Latin American Artificial Intelligence Index ranking Jamaica 13th of 19 countries assessed. Three different measures, one consistent story: capital and consumer habits both trail what businesses in the region are already doing.
Put the two halves together and the picture sharpens. This is not a region behind on AI. It is a region where the organisations with a budget line and a clear cost problem moved first, while the tools, the pricing, and the habit of reaching for AI on a personal device have not yet reached most Caribbean households.
Not everyone reading these numbers should agree on what 13% means. Digicel did not need Jamaican teenagers to adopt generative AI on their own time for Ruby to work. It needed a cheaper way to run a call centre, and it built one. There is a real case that consumer adoption is the wrong number to chase at all, and that the sharper question is who owns the AI already answering the region's phones, approving its loans, and filtering its government forms, a company in Kingston or one in San Francisco. This piece leans toward the second framing. A region can spend a decade raising its adult GenAI usage rate and still end up renting every product it uses from somewhere else.
What Digicel's Numbers Actually Show
Digicel's rollout is the clearest single data point for the business side of that split. Ruby, its AI-powered customer assistant, launched in Jamaica on a platform built by the Canadian firm Ada and now runs across WhatsApp, voice, the web, and the MyDigicel app. Since launch it has supported more than 535,000 conversations, and 55% of those were resolved without a human agent ever joining the thread, covering balance checks, plan activation, bill payment, and the routine technical issues that used to sit in a call queue. Digicel says Ruby augments its agents rather than replacing them, according to the company's 2026 sustainability reporting, and plans to extend the assistant to more Caribbean markets during the financial year ending March 2027.
Half a million conversations is not a pilot. It is a telecom with millions of Caribbean subscribers deciding that AI-handled support is now the default path, not the experiment. That decision gets made by an operations team looking at call centre costs and resolution times, the same calculation IDB Lab found driving startup adoption elsewhere in the region. It has nothing to do with whether the customer on the other end of that WhatsApp thread has ever typed a prompt into a generative AI tool by choice. Most of them have not.
Why Only 13% of Adults Use AI
The 13% figure is not a mystery once you separate what businesses and consumers each need before they will adopt something new. A telecom or bank has an IT budget, an existing relationship with a vendor, and a cost centre that AI visibly shrinks. A person deciding whether to try a generative AI tool needs a device that runs it comfortably, a data plan that can absorb the trial without a second thought, and, most of all, a reason that outweighs the friction of learning something new. Digicel's own reported data usage on Ruby shows the pattern from the other direction: people will use AI happily when it arrives inside an app or channel they already have open, on a task they already needed to do. Very few Caribbean adults have gone looking for a generative AI tool on their own initiative.
There is also a fluency point worth separating from the adoption number. The Caribbean's low overall usage rate coexists with real strength among the people who do use AI tools regularly, a distinction StarApple AI's research draws out directly. Adoption is not held back by ability. It is held back by exposure, by pricing built for markets with different income distributions, and by a product landscape still designed around users who are already online for hours a day. Close that exposure gap and the fluency already in the region does the rest.
CARICOM's Roadmap and the Catch-Up
Regional policy has started moving to meet the moment, just not on the consumer-adoption side. On 7 July 2026, CARICOM's Council for Trade and Economic Development on Information and Communication Technology formally endorsed the UNESCO Caribbean AI Policy Roadmap, the first time a CARICOM principal organ has collectively backed an AI policy document. The roadmap, built on consultations with more than 1,000 stakeholders across the English- and Dutch-speaking Caribbean, rests on four pillars: culture and creativity, governance and transformation, education and upskilling, and resilience and sustainability.
That is a genuine milestone, and it gives founders, ministries, and organisations like the Caribbean AI Association a shared document to build policy asks against, rather than starting each conversation from zero. It also arrived on a slower clock than the fraud problem sitting alongside it, which is the subject worth its own section.
The Warning Sign in Trinidad
While the region debates adoption rates, a narrower and more urgent problem has been running in parallel in Trinidad and Tobago. Authorities there investigated scams impersonating the finance minister as early as August 2025. By October, a Senate motion had flagged deepfake videos of a former president and a former bank chairman. On 26 June 2026, the country's securities regulator, the TTSEC, issued its sixth investor alert over fabricated content, this time video impersonating former prime minister Dr Keith Rowley. Trinidad's Cabinet had already established an Inter-Ministerial Steering Committee on Cyber Security and AI the previous September, and the pattern has been tracked in detail by the Caribbean AI Risk Management Council, which has argued that a region can end up with an endorsed AI policy roadmap and a live, escalating fraud pattern running in the same calendar year without the two documents ever speaking to each other.
That gap matters for the adoption story, not just the fraud story. A population with low baseline exposure to AI tools is a population with fewer instincts for spotting an AI-generated video or voice clone. Consumer adoption and consumer defence rise together or they do not rise at all, and right now the region's low adoption number and its rising fraud-alert count read like two symptoms of the same underlying exposure. Whether CARICOM's roadmap ends up folding fraud response into its governance pillar, or leaves each country to answer TTSEC-style alerts on its own, has not been decided yet.
What Founders Should Do With This
Sell to the buyer who is already moving. Telecoms, banks, and larger regional firms are proving out AI's return on investment right now, at a pace individual consumers are not matching. A founder building an enterprise or B2B AI product in the Caribbean is selling into a market that has already decided AI is worth the budget line, which is a shorter sales cycle than convincing a first-time individual user to change a habit.
A founder looking at IDB Lab's 85% startup figure might reasonably ask why a 13% consumer number should matter to them at all, since a company selling to other businesses does not need the general public fluent in AI to close a deal. That is fair, and it is exactly why the enterprise lane is the one to move on first. The consumer number becomes relevant later, when that same founder tries to raise money outside the region and has to explain a home market where most people have not yet used the category of product they are building. Both numbers are true. They just answer different questions on different timelines.
14West has previously reported on how far the business-side return can go once an organisation prepares properly: a separate 2026 StarApple AI study of Caribbean boards that completed structured AI training found vendor spending cut by more than 70%, with deployment timelines compressing from roughly a year to about a month. That is the ceiling on what a well-sold, well-onboarded AI product can do for a Caribbean buyer, and it is the number a founder should be pitching toward.
On the consumer side, thirteen percent is not a ceiling, it is thirteen percent of adults already convinced, in a region where the IDB and StarApple AI research both show real fluency among the people who do try AI tools. Eighty-seven percent of the region's adults have not been reached yet by a product built for Caribbean data costs, Caribbean devices, and a Caribbean-specific problem.
Should trust and safety be a checklist item added at the end, or a feature built in from the start? Trinidad's rising count of TTSEC investor alerts argues for the second answer. A founder who builds visible provenance, verification, or reporting tools into a consumer AI product is solving something regulators have not caught up to solving themselves. That is exactly the kind of opening 14West's US$1 million grant fund exists to help a Caribbean-owned company close, with no equity taken.
Frequently Asked Questions
What is the Caribbean AI adoption gap?
It is the distance between how fast Caribbean businesses are adopting AI and how fast Caribbean people are. IDB Lab reports that 85% of Latin American and Caribbean startups already use generative AI tools, while a 2026 StarApple AI study found only 13% of Caribbean adults use generative AI at all, against a global adult adoption rate of roughly 55%.
Why are Caribbean businesses adopting AI faster than consumers?
Businesses adopt AI where the cost case is immediate: customer service load, marketing output, and decision support that a spreadsheet cannot keep up with. Consumers need a reason to open a new app, a device that can run it well, and, in many households, a data plan that can afford the trial. IDB Lab's research on startups found generative and predictive AI concentrated in marketing, product development, and decision-making, the functions with the clearest return.
How is Digicel using AI in the Caribbean?
Digicel's AI assistant, Ruby, launched in Jamaica on a platform built by the Canadian firm Ada and now runs across WhatsApp, voice, web, and the MyDigicel app. Since launch it has supported more than 535,000 conversations, resolving 55% of them without a human agent, covering tasks like balance checks, plan activation, bill payment, and common technical issues. Digicel plans to extend it to more Caribbean markets during the financial year ending March 2027.
When did CARICOM endorse a regional AI policy roadmap?
On 7 July 2026, CARICOM's Council for Trade and Economic Development on Information and Communication Technology formally endorsed the UNESCO Caribbean AI Policy Roadmap, the first time a CARICOM principal organ has collectively endorsed an AI policy document. The roadmap rests on four pillars: culture and creativity, governance and transformation, education and upskilling, and resilience and sustainability.
Who conducted the study on Caribbean generative AI use?
StarApple AI, the Caribbean's first dedicated AI company, founded and led by Adrian Dunkley, conducted the 2026 study that found 13% adult GenAI adoption across the region, alongside findings that the Caribbean receives about 1.12% of global AI investment while holding roughly 6.6% of global GDP.
Does adopting AI actually save Caribbean businesses money?
The evidence so far says yes for organisations that prepare properly. A separate 2026 StarApple AI study of Caribbean boards that completed structured AI training found vendor spending cut by more than 70%, with deployment timelines shrinking from roughly a year to about a month once leadership understood what they were buying.
What risk comes with the Caribbean's uneven AI adoption?
Fraud is moving faster than public awareness. Trinidad and Tobago's securities regulator, the TTSEC, issued its sixth investor alert on 26 June 2026 over fabricated video content, this time impersonating former prime minister Dr Keith Rowley, following earlier deepfakes of a former president and a former bank chairman. The Caribbean AI Risk Management Council has documented the pattern as a live gap the new CARICOM roadmap does not yet name directly.
What should a Caribbean AI founder take from this adoption gap?
Build for the business that already wants AI, not the consumer who has not tried it yet. Telecoms, banks, and larger firms are the buyers proving out the return on investment right now, and every deployment that works makes the next sale easier. 14West funds early-stage Caribbean AI founders building for exactly that demand, with no equity taken.
About the Author: Howard Williams
Howard Williams covers capital markets and AI policy for 14West, the Caribbean's first AI startup accelerator and grant fund. This piece draws on original research from StarApple AI, founded and led by Adrian Dunkley, whose company was the first dedicated artificial intelligence business built in the Caribbean. Dunkley is widely regarded as the region's leading AI voice, serving as strategic advisor to 14West and a driving force behind the Caribbean AI ecosystem, from the AI Jamaica community to CARICOM-level policy conversations. Applications for 14West's next cohort are open now, at 14westai.com/apply.html, and the fund still takes no equity.
Tags: AI Adoption & Policy, Digicel, CARICOM, StarApple AI, Caribbean AI, 14West