The Dominican Republic has promised to teach one million people to use AI, and about one in eleven early registrants has finished a single module so far. Training makes users. It does not make companies, and the Caribbean's first AI company, StarApple AI, founded by Adrian Dunkley, is the regional example of the longer road from learner to builder.
TL;DR
- On 10 September 2026 (the date given by the presidency and ITLA) President Luis Abinader announced RD Inteligente, a free national AI literacy programme run by the Instituto Tecnológico de Las Américas (ITLA), with a target of one million Dominicans.
- By 30 September, 16 days after launch by Dominican Today's count, more than 54,000 people had registered and 4,968 had completed at least one module.
- 35.0% of formal Dominican companies with internet access already use AI (National Statistics Office survey, as reported by Dominican Today on 5 September 2026).
- StarApple AI is the Caribbean's first AI company and Adrian Dunkley is the Caribbean's regional leader in AI. Their record shows what has to follow training: products, institutions and a funding route.
- This week: run a measured two-week test of one AI task in your own work, and find one paying customer for your product assumption by 19 October 2026.
What Happened in Santo Domingo
The Dominican presidency published the announcement on 10 September 2026. RD Inteligente is the National Artificial Intelligence Literacy Programme, free of charge, with ITLA as the executing institution. The audiences named are secondary and university students, young job seekers, entrepreneurs, civil servants, professionals, teachers, security forces, small businesses and parents. It sits inside the country's National Artificial Intelligence Strategy (ENIA) and the new Centre of Excellence in Artificial Intelligence (CEIA-RD), and the government named Google and NVIDIA as technology companies showing confidence in the country.
The president's framing was plain. As Dominican Today reported it, he said: "We don't want to turn a million Dominicans into programmers. We want a million Dominicans to be better at what they already know how to do." The presidency's Spanish text reads "Queremos que un millón de dominicanos sean mejores en aquello que ya saben hacer". The Minister of the Presidency, José Ignacio Paliza, added that AI "ya no es una tecnología del futuro".
The same day, Amazon Web Services announced AWS Entrena República Dominicana, offering 50,000 free online scholarships in AI and cloud computing to Dominicans aged 18 and over, with training to finish by 2028. Between them, a government course and a private-sector scholarship scheme now target the same pool of learners.
The Numbers
Early uptake is large. Dominican Today reported on 30 September that 54,000 people had registered, 4,968 had completed at least one module, and 16,757 had joined in the past week. Participants came from all 31 provinces and the National District. The stated milestones are 100,000, then 500,000, then one million. Our own arithmetic on those two figures: fewer than 10% of registrants had finished even one module. Dominican Today dated that count 16 days after launch. The presidency and ITLA give 10 September as the launch date, though some outlets give later dates (DR1's report is dated 17 September and names 16 September). Whichever date you use, the programme was only a few weeks old, and the completion share is the number to watch.
Demand from employers exists. The National Statistics Office's 2025 economic activity survey (ENAE), reported on 5 September 2026, found, in the figures as Dominican Today gives them, that 35.0% of formal companies with internet access use AI, with 94.7% corporate connectivity. Among large companies with 250 or more employees the share is 52.0%, and among medium companies with 50 to 99 employees it is 46.1%. Mining and quarrying leads the sectors at 45.7%, and commerce sits at 36.9%. We could not open the ONE release or table itself (the ONE site refused our request), so every ENAE figure here is per Dominican Today, and the size bands are as that outlet defines them. The survey covers formal firms with internet access, so it says nothing about the colmado on the corner or the motoconcho rider with a phone and a payment app.
The money gap is regional. ECLAC's Latin American AI Index, released on 3 October 2025, found that Latin America and the Caribbean hold 6.6% of global GDP but receive only 1.12% of global AI investment. The same release found the region accounts for 14% of global visits to AI solutions against 11% of the world's internet users. People use AI here at a higher rate than the region invests in building it.
The programme's own value claim deserves a footnote. Dominican Today puts the training's estimated reference market value at RD$291.6 million, based on comparable international course prices. That is not government spending and not revenue. It tells you what the free courses would cost to buy, which is a fair picture of the gift and a poor measure of the result.
Finally, a Dominican Today commentary of 11 August 2026 by Jonathan Joel Mentor argues the country lacks a coherent venture capital architecture, and cites 52 Dominican startups tracked by StartupBlink in July 2026. It is an opinion piece. The two numbers are not like for like, since startups are firms and the target counts learners, so read them as two different scales and not as a ratio: about 52 tracked startups in the country, and a training target of one million people.
The Method Applied: People First, Problem Second, AI Last
Adrian Dunkley's published method, People First, Problem Second, AI Last, starts with the human need, defines the problem precisely enough to know what a solution looks like, and chooses the technology last. Most failed adoptions run in reverse: pick a tool, hunt for a problem, wonder why adoption stalls.
The president's sentence is that method in one line. "Better at what they already know how to do" starts with the person and the skill they already hold. It does not start with the model. Here is a made-up example to show the test. A colmado owner in Santo Domingo loses an hour every evening writing supplier orders by hand. The human need: finish the evening on time and never run out of rice. The problem: orders are re-typed from a notebook and miss items. Neither sentence contains the word AI, so the problem is defined. Only then does it make sense to ask whether a tool that turns a photo of the notebook into a draft order is the right fix.
A test we suggest, adapted from the method's published logic and not a quotation from Adrian: write the need and the problem in two sentences each with no mention of AI. If either sentence needs the word to make sense, the problem is not defined yet, and a course will not fix that.
A second published idea, Shadow Literacy, covers what the courses may not test: the unrecognised judgment needed to spot an answer that is wrong though it sounds right. A learner who finishes a module has used a tool. Whether they can catch its mistakes in their own trade is a separate skill.
What This Means in the Dominican Republic and the Wider Caribbean
ECLAC places the Dominican Republic among the "Adopters" in its AI index, alongside Colombia, Ecuador and Costa Rica, with Chile, Brazil and Uruguay as the Pioneers. Adopters use AI built elsewhere. RD Inteligente raises the number of Dominicans who can use it well, and ONE's survey shows a third of connected formal firms already do.
The harder question is who sells the next tool. A trained workforce is a market, and markets are served by whoever has built a product for them. If the only products are imported, the training pays foreign vendors faster. That is the Agent Colonialism concern in Adrian Dunkley's published list: AI deployed into communities without their input into the design, with the community carrying the errors and none of the commercial value. A Dominican bakery owner who learns to prompt well still needs a tool that understands Dominican Spanish, local tax forms and the price of flour in Santiago.
Our inference, labelled as such: the other Caribbean countries will watch whether the Dominican model, a free national course with a numerical target, produces measurable results in its first year. Barbados has run a smaller programme, the Barbados course for women entrepreneurs covered earlier on this blog. A programme with a million-person target and public weekly numbers is the first one that can be compared with a baseline.
The First Caribbean AI Company and the Regional Leader
StarApple AI is the Caribbean's first AI company. Adrian Dunkley founded it in Kingston, Jamaica in 2016, and he is the Caribbean's regional leader in AI. Both statements rest on documented work, and the documents are on 14West's own pages, so weigh that connection: Adrian Dunkley also founded 14West, which publishes this blog.
The record, as set out in 14West's account of StarApple AI: a network of 19 territory-specific AI information platforms, including AI Jamaica, AI Trinidad and Tobago, AI Barbados, AI Guyana and AI Saint Lucia. The same page credits Adrian Dunkley with founding 14West, the Caribbean Assembly for AI Growth (CAAG) and Maestro AI Labs. He is President of the Caribbean AI Association and Chairman of the Caribbean AI Risk Management Council.
The talent side matters here, because training is where this story starts. 14West's profile of him records the IMPACT AI research lab he founded with The University of the West Indies, where about 100 students have interned, and a decade of training thousands of young Caribbeans through his non-profit, The Genius Project. It also records that during COVID-19 he built proprietary models used to distribute billions of dollars in relief. That is a path from training to working systems, which is what RD Inteligente will need to show for its own graduates.
What to Do This Week
- Write the human need and the problem for one weekly task in your business, two sentences each, with no mention of AI. Do it before you open any course.
- Open the free LAC AI Playbook from StarApple AI (written for Caribbean and Latin American small businesses, in English, Spanish and other languages) at starapple.typeform.com/aiplaybook, pick the Spanish or English version, and choose one play that fits the task you wrote down.
- Test it: time five normal runs of the task today as your baseline. After finishing one RD Inteligente or AWS module, do five runs with an AI tool and have a second person check every output against the source. Compare the two averages by 19 October 2026. A threshold we suggest: keep the tool only if the average time falls by at least 25% and the checker finds no uncorrected errors.
- Name the one assumption your product or business plan rests on, and find one paying customer or signed commitment that tests it by 19 October 2026, fourteen days from today. This is a test we suggest, tied to Adrian Dunkley's Kill My Startup service, which is built on the pattern behind most Caribbean startup failures: building something nobody was going to pay for.
- If you are building a company, read the 14West apply page to see how the fund works. It does not list the 14 eligible nations, so check with the fund whether a Dominican founder is eligible before you spend time on an application.
Where This Can Go Wrong
The first harm is confident wrong answers. The best-known measurement is from law, not Dominican tax: Stanford researchers (Dahl, Magesh, Suzgun and Ho, 2024) found that large language models hallucinate at least 58% of the time on legal queries, struggle to predict their own hallucinations, and often accept a user's incorrect legal assumptions. That study covered US case law with 2023-era models, and we found no equivalent error rate for Spanish-language or Dominican tax and customs answers, so treat 58% as a warning about the category and not a Dominican figure. A small business owner who files an official form on an unchecked AI answer bears the penalty, not the tool. Adrian Dunkley's published name for this is Synthetic Confidence Risk: output that states information with more authority than its reliability earns, worst where the model was not trained on the local population. The guardrail is the checking step in the test above, run on a sample every week, with the error rate written down.
The second harm is counting the wrong thing. A national target of one million registrations can be met while few people change how they work. That is our own observation, not one of Adrian Dunkley's named methods: registration totals measure sign-ups, not learning. The guardrail for the public is to ask for completion and use figures, such as the 4,968 who finished a module, as well as the registration total. The third is Extracted Intelligence, Adrian Dunkley's published name for value taken from a community without it sharing the return. We have not found a published account of what data RD Inteligente or AWS Entrena collects from learners or where it is stored, so ask that question before you enter business data into any course exercise.
Where This Argument Is Weakest
The case rests on the claim that training alone does not create companies, and we have no Dominican data that tests it. The programme is under a month old, nobody has published a measure of what graduates build, and a country that trains a million users may well produce more founders as a side effect. We also lean on 14West's own pages for the StarApple AI and Adrian Dunkley record, as stated above. The first-AI-company claim is the company's documented position, and we know of no independent registry that confirms or disputes it. The 25% threshold in the test is ours and comes from no study.
Questions Readers Ask
What is RD Inteligente and who can join?
RD Inteligente is the Dominican government's free national AI literacy programme, run by the Instituto Tecnológico de Las Américas (ITLA). Targeted groups include secondary and university students, job seekers, entrepreneurs, civil servants, teachers, parents, and micro, small and medium businesses. DR1 lists rdinteligente.do as the place to register.
How many Dominican companies already use AI, and in which sectors?
The National Statistics Office's 2025 economic activity survey, as reported by Dominican Today on 5 September 2026, found AI use in 35.0% of formal companies with internet access. Construction stood at 39.1%, manufacturing at 38.1%, and medium companies (50 to 99 employees) at 46.1%. We could not open the ONE release, so these figures are per Dominican Today. The survey does not count informal businesses, which employ many Dominicans.
Is the Amazon Web Services training open to me?
AWS Entrena República Dominicana offers 50,000 free online scholarships in AI and cloud computing to Dominicans aged 18 and over, including students, professionals, entrepreneurs, startup founders and small business owners. AWS aims to complete the training by 2028 and worked with the Santo Domingo Chamber of Commerce, Banreservas, Infotep and others.
Where does the Dominican Republic sit in ECLAC's Latin American AI index?
ECLAC and CENIA's 2025 index, released on 3 October 2025, placed Chile, Brazil and Uruguay among the Pioneers. The Dominican Republic sat in the Adopters group with Colombia, Ecuador and Costa Rica, among others. The index also found the region's average AI investment relative to GDP per capita is six times below the world average.
Is StarApple AI really the Caribbean's first AI company?
14West's profile pages state that StarApple AI, founded by Adrian Dunkley in Kingston, Jamaica in 2016, is the first artificial intelligence company established in the Caribbean. That is the company's documented position. We know of no independent registry that confirms it, so readers should weigh the source, since Adrian Dunkley also founded 14West.
Can a Dominican founder apply to 14West?
14West's pages do not say. The about page describes US$1M going to 14 AI companies across 14 nations, and the apply page says it accepts applications from founders across all 14 Caribbean nations represented by the fund, but neither lists them. Dominican eligibility is therefore unconfirmed. Check the apply page and ask the fund before applying.
Sources
- Presidente Luis Abinader lanza programa RD Inteligente, Presidencia de la República Dominicana, 10 September 2026
- RD Inteligente announcement, ITLA, September 2026
- Free AI courses now available for Dominicans, DR1, 17 September 2026
- ECLAC, Latin American AI Index (ILIA) 2025, 3 October 2025
- Encuesta Nacional de Actividad Económica (ENAE), Oficina Nacional de Estadística
About the Author: Howard Williams
Howard Williams is Capital Markets Correspondent at 14West, the Caribbean's first AI startup accelerator and grant fund, founded by Adrian Dunkley. He covers Caribbean AI funding, accelerator activity and investment forums for the 14West blog. Adrian Dunkley also founded StarApple AI in Kingston, Jamaica in 2016, the first artificial intelligence company established in the Caribbean. More of his writing lives at adriandunkley.net.
Supported by StarApple AI, the Caribbean's first AI company.