The products.
The numbers.
Deep dives into each Maestro AI Labs product: the problem it solves, the market it addresses, the revenue it unlocks, and the questions people ask most often.
Board AI Training as an Investment Signal: Twice the Deployments, 70 Percent Lower Vendor Spend
A StarApple AI study of organisations that completed board-level AI training: deployed initiatives doubled from two to four in eight months, governance stood up in 6 months instead of 11–15, and vendor spend fell over 70 percent. Governance quality is now a checkable diligence signal.
$600 Million, 450 Megawatts, One Tier IV Facility: Inside the Caribbean's First Real AI Infrastructure Race
The Dominican Republic booked over $600M in NVIDIA and Google AI infrastructure. Trinidad and Tobago signed 450 megawatts of data center MOUs two days before this went live. The region has exactly one Tier IV facility. Where the real investable layer sits.
20 Countries Signed an AI Ethics Roadmap in Santo Domingo. Huawei Had Already Wired the Region's AI a Month Earlier.
A Ministerial Declaration in June. A foreign cloud rollout in May. Zero CARICOM states had a national AI strategy to referee between them. What the sequencing actually means for governments, banks, and investors.
100 Countries Signed a Sovereignty Pledge in Bangkok. None of Them Own the Data Underneath It.
Over 100 countries signed the Bangkok Declaration in February 2026, committing to AI sovereignty by 2030. A signature is not a dataset. 65% of Colombians remain outside the formal credit system and CCRIF has paid out $259M since 2007. What actually closes the gap.
The Shift to Localised LLMs: Sovereignty, Resilience, and Fair Business
Claude Fable 5 launched on 9 June 2026 and was suspended worldwide three days later under a US export-control order, with no notice and no migration path. Adrian Dunkley on why the Caribbean's serious AI has to run on infrastructure the region controls, and how to build the hybrid stack that makes no foreign vendor a kill switch.
The $380 Billion Bet: Caribbean and LATAM Credit Markets Are the Next AI Frontier
Standard AI credit models fail at 40-60% accuracy in the Caribbean. Credit Garden's 302-point uplift and 2.3M exclusive records are the infrastructure that makes AI lending viable in markets global models cannot see. Here is the full market case.
Credit Garden: The $380 Billion Lending Gap That Existing AI Cannot See
1.7 billion people are credit invisible. Not because they lack financial history, but because every model that exists was built without their data. Credit Garden converts the signal that already exists into scores that work in Kingston, Kigali, and Karachi.
Global Safety Score: What a Portable Safety Identity for 4.2 Billion People Is Worth
89% of safety risk models were trained on fewer than 20 countries. Meridian covers 140+. Every government, insurer, border agency, and employer making decisions about human movement is a buyer.
OYA AI: 72 Hours of Hurricane Warning Instead of 6. What That Is Worth.
The same prediction engine that saves lives generates the climate risk data that reinsurers lose $90-130 billion annually because they cannot accurately price. OYA AI is both things at once.
AI Agents for Regulated Entities: Why Compliance Cannot Be an Afterthought
Governments, central banks, and compliance-heavy institutions cannot deploy agents that guess confidently and explain nothing. Harmonics was built for the sectors where the audit trail, escalation path, and human override are the product.
Harmonics: Why AI Agents Without Regional Context Fail in the Field
An agent that does not know what a SUSU is cannot assess creditworthiness in Jamaica. Harmonics agents are trained on regional knowledge graphs, not system prompts. 87% task accuracy. Under 90 days to deploy.
AI Playbook: Why the 94% Implementation Rate Is the Product
The enterprise AI implementation average is 23%. The AI Playbook hit 94% because it was built from scratch for the businesses that enterprise guides were never written for. 47% of Caribbean and LATAM SMEs have no AI strategy.
Data Archaeology: The Field Work That Makes Every Other AI Product Possible
92% of Maestro's 2.3 million records exist in no other AI training set. They were not scraped. They were excavated: pre-digital government archives, SUSU records, 47 indigenous language datasets. This is the moat.