- Hurricane Melissa struck Jamaica as a Category 5 storm on 28 October 2025, the most intense hurricane on record to hit the island, causing an estimated US$12.2 billion in total damage and losses, equal to 56.7% of Jamaica's 2024 GDP, per the Planning Institute of Jamaica.
- CCRIF SPC paid the Jamaican government US$91.9 million within 14 days, the largest combined payout in the facility's history, because its parametric policies trigger on measured wind speed and rainfall, not on inspecting a single property.
- Verisk, Moody's RMS and Cotality put insured property losses somewhere between roughly US$1 billion and US$5 billion, and the Insurance Association of Jamaica says fewer than one in five homes on the island carry any coverage at all.
- Seven months after landfall, Jamaica Observer reporting found individual homeowner claims still open, even though Jamaica's Insurance Regulations require settlement within 30 days of the conditions for payment being met.
- The gap between a 14-day sovereign payout and a seven-month personal claim is a data problem before it is a fairness problem, and it is the exact gap Maestro AI Labs built Global Safety Score and Data Archaeology to close.
Fourteen days after Hurricane Melissa made landfall near New Hope, Westmoreland, with sustained winds of 185 miles an hour, CCRIF SPC had already transferred US$70.8 million to the Government of Jamaica under its tropical cyclone policy, the largest single payout the regional catastrophe facility had made in its history. A second payout of US$21.1 million followed days later under the government's excess rainfall policy, bringing the combined total to US$91.9 million. Seven months on, a Jamaica Observer op-ed described a homeowner's claim as sitting exactly where it had sat in November: open, unresolved, and effectively in the dark. Both facts describe the same storm. They describe two different insurance systems responding to it at two very different speeds.
What Melissa Actually Cost
Melissa made landfall as a Category 5 hurricane on 28 October 2025, the most intense storm ever recorded to strike Jamaica. The Planning Institute of Jamaica puts total damage and losses at J$1.952 trillion, roughly US$12.2 billion, equivalent to 56.7% of the country's 2024 gross domestic product. AccuWeather's broader economic-loss estimate, which factors in lost output and long-term disruption rather than physical damage alone, runs as high as US$48 billion to US$52 billion. Westmoreland, where the storm came ashore, lost thousands of structures outright.
Set against that, the private insurance market's estimate of what it actually owes is a fraction of the total. Verisk's Extreme Event Solutions group puts insured onshore property losses between US$2.2 billion and US$4.2 billion. Moody's RMS estimates a range of US$3 billion to US$5 billion, with a best estimate around US$3.5 billion. Cotality's early estimate sat lower still, between US$1 billion and US$2.5 billion. The spread between those three firms is itself a signal: modellers with access to the same storm data are producing loss ranges that differ by billions, because none of them has a complete, current picture of what stood on the ground in Westmoreland, St Elizabeth or Manchester the night before landfall.
The reason the modelled range is so wide traces back to a single, well-documented number. The Insurance Association of Jamaica reports that fewer than one in five homes on the island carry any insurance coverage, and many of the policies that do exist are underinsured relative to actual rebuild cost. A market where 80% of the housing stock has never been priced by an insurer is a market an insurer cannot model with much confidence after the fact either.
The 14-Day Payout That Actually Worked
CCRIF's speed is a design choice, not a customer-service achievement. The facility's tropical cyclone and excess rainfall policies are parametric: they pay a pre-agreed amount once an independently measured trigger, wind speed at a set of grid points, or accumulated rainfall over a defined window, crosses a threshold written into the contract years in advance. Nobody at CCRIF needed to send an adjuster to a single roof in Westmoreland to release the US$70.8 million wind payout or the US$21.1 million rainfall payout. The model runs on meteorological data that exists the moment the storm has passed, not on a property inventory that has to be built claim by claim.
That design is also why the payout landed with the government rather than with individual households. CCRIF's client is the state, and the money is meant to fund immediate liquidity, emergency services, shelter, the first weeks of a response before slower-moving reconstruction financing arrives. It arrived alongside other fast-moving instruments: a World Bank catastrophe bond, renewed in May 2024 for US$150 million to cover the 2024 through 2027 hurricane seasons, was fully triggered by Melissa. The IMF approved a US$415 million disbursement in January 2026 specifically for hurricane relief, part of a wider financing framework of up to US$1.077 billion in combined government and multilateral resources. Jamaica went on to secure a US$6.7 billion recovery and reconstruction plan by December 2025. At the sovereign level, the machinery Jamaica had built before the storm did exactly what it was built to do.
The Seven Months Nobody Budgeted For
Individual claims run on a different mechanism entirely, and it is the one that stalled. Indemnity insurance, the kind a homeowner buys on a house, does not pay against a wind-speed threshold. It pays against an assessed loss: what the property was worth before the storm, what it will actually cost to repair, verified against a policy an adjuster has to interpret line by line. That verification depends on a property record, and in most of Jamaica, that record barely exists in a form an insurer or a modeller can use. Title history is incomplete. Construction detail, roof material, elevation, prior claims, is not digitised. Flood and wind-zone maps have not been rebuilt at the resolution modern climate risk actually requires.
The Jamaica Observer's June 2026 reporting on the aftermath found policyholders whose claims had been open for months without a resolution, and pointed to a specific regulatory failure behind the delay. Jamaica's Insurance Regulations 135 require insurers to settle a valid claim within 30 days of the conditions for payment being met, with statutory interest running from that point. The Financial Services Commission's 2022 Market Conduct Rules separately require settlement to be handled fairly and without undue delay. Both rules exist. Neither, according to that reporting, has been enforced with any consistency: market conduct has been treated as an aspiration rather than a supervisory priority for roughly two decades, and there is no independent insurance ombudsman in Jamaica with the authority to issue a binding ruling in a policyholder's favour. A homeowner whose roof is gone and whose insurer has gone quiet has very little standing to push back, and, as the Observer put it, most affected policyholders have no public platform and cannot afford to make an enemy of the only company they may ever need to renew a policy with again.
| Figure | Source | Status |
|---|---|---|
| $12.2B total damage; 56.7% of 2024 GDP | Planning Institute of Jamaica, via Jamaica Information Service | Independent, government agency |
| $91.9M CCRIF payout in 14 days ($70.8M wind, $21.1M rainfall) | CCRIF SPC official payout announcements | Independent, regional facility |
| Fewer than 1 in 5 homes insured | Insurance Association of Jamaica | Independent, industry body |
| $2.2B-$4.2B / $3B-$5B / $1B-$2.5B insured loss ranges | Verisk, Moody's RMS, Cotality (respectively) | Independent, catastrophe modellers |
| 2026 Atlantic season: 8-14 named storms; Caribbean sea temps 27.5-30°C | NOAA 2026 seasonal outlook | Independent, US federal agency |
| 140+ countries covered; 89% of comparable models cover fewer than 20 | Maestro AI Labs, Global Safety Score product data | Company-reported, not independently audited |
Two Different Problems Wearing One Insurance Label
It is tempting to read the CCRIF payout and the stalled homeowner claims as the same story told twice, one fast, one slow, both about Hurricane Melissa. They are not the same story. A parametric trigger and an indemnity claim solve for different questions with different data, and conflating them is how a region ends up congratulating itself on disaster response speed at the sovereign level while individual households go seven months without an answer. CCRIF answers the question "did a storm of this intensity occur here." An adjuster has to answer "what, specifically, did this storm do to this specific house," and that second question cannot be answered from satellite wind data. It requires a property record that was accurate before the storm and a damage assessment that can be verified after it, at a resolution most Caribbean insurance markets have never built.
That resolution gap is the actual product opportunity, and it is why Maestro AI Labs exists as a second company alongside StarApple AI rather than as a feature bolted onto one. StarApple AI, founded in Kingston in 2023 as the first AI company established in the Caribbean, built much of the underlying research into how thin the region's structured data actually is. Maestro AI Labs' Global Safety Score product, internally called Meridian, covers more than 140 countries on portable risk and safety identity; the company reports that 89% of comparable safety models in the market cover fewer than 20 countries, which is roughly the scale problem Caribbean insurers hit the moment they try to underwrite a market most global risk models were never built to see. Data Archaeology, a separate Maestro AI Labs product, exists to reconstruct exactly the kind of property and asset history that an indemnity claim in Westmoreland actually needs and that no single Jamaican insurer has ever had reason to digitise on its own.
"CCRIF proved something important: when the data a model needs already exists, in this case wind speed and rainfall measured from space, the Caribbean can move faster on disaster payouts than almost anywhere in the world. Fourteen days is a genuinely good number. The problem is that almost none of the data an individual claim needs looks anything like that. Roof type, elevation, prior repairs, none of it is sitting in a satellite feed. Somebody has to build that layer, or every storm after this one repeats the same split screen: a government paid in two weeks, a homeowner still waiting at Christmas."
Adrian Dunkley, Co-Founder, Maestro AI Labs
Why This Matters Before the Next Storm, Not After It
The 2026 Atlantic hurricane season is past its midpoint and inside its highest-risk window as this is written. NOAA's official outlook, issued in August, gives a 55% chance of a below-average season by storm count: 8 to 14 named storms, 3 to 6 hurricanes, 1 to 3 major hurricanes. A below-average count has rarely translated into a below-average loss year in the Caribbean, and the reason is sitting in the ocean itself. Sea surface temperatures across the Caribbean basin are running between 27.5°C and 30°C, warm enough to support the kind of rapid intensification that turned Melissa from a tropical system into a Category 5 hurricane in a matter of days. A season with fewer storms can still produce one storm that behaves exactly like Melissa did, and the region's response mechanism would split the same way: fast at the sovereign level, slow at the household level, for the same structural reason.
None of this argues that CCRIF, the World Bank bond, or the IMF disbursement did anything wrong. They performed exactly as designed, and Jamaica is materially better off for having built that layer of sovereign risk transfer before Melissa arrived than it would have been without it. The argument is narrower and more specific: sovereign-level speed was never going to trickle down to individual claims on its own, because the two systems solve different problems with different inputs, and only one of those input sets has been built out at Caribbean scale.
- Regulators need enforcement teeth, not new rules. Jamaica's 30-day settlement window and its 2022 market conduct rules already exist on paper; what is missing, per the Observer's reporting, is a binding ombudsman function and a supervisory culture that treats consumer protection as seriously as solvency.
- Insurers need a property record they can actually underwrite against. Roof material, elevation, construction type and claims history have to exist in a usable, current form before the next storm, not reconstructed from memory and photographs after it.
- Households can start documenting now, at no cost. A dated photo inventory of a home's structure and contents, taken before hurricane season peaks each year, is the single cheapest thing a Caribbean homeowner can do to shorten their own claim later.
- Capital and data providers have a genuine opening. A region that can move US$91.9 million in 14 days at the sovereign level, but needs seven months and counting to resolve a single roof claim, has plenty of urgency and capital appetite; what it lacks is the data layer that turns speed at one level into speed at the other. That is the specific gap Maestro AI Labs built Global Safety Score and Data Archaeology to close, and the reason Adrian Dunkley has spent more than a decade arguing that Caribbean AI has to be built on Caribbean data rather than borrowed from models trained somewhere else.
Caribbean insurers, regulators and investors who want to talk through what a property-level data layer would actually take to build can write to Maestro AI Labs at ceo@maestrosai.com. More on Adrian Dunkley's broader case for Caribbean-built AI infrastructure is at adriandunkley.net.