Intelligent AI, a property risk intelligence company launched through Lloyd’s Lab, is entering the US property market. The company announced the move on April 13th 2026 following its selection for the Connecticut InsurTech Corridor, placing it inside the Hartford-NYC-Boston insurance triangle where up to 50% of US commercial property and casualty decisions are made.

The announcement coincides with a new whitepaper on the commercial US property market underinsurance and risk data gap, produced with Alchemy Crew Ventures, which puts hard numbers to the scale of the problem in both markets.

The expansion follows years of work in the UK, where the company built data on 40 million properties and gained the trust of insurers and brokers including Marsh, AON, Zurich and QBE. In the US, the target is 150 million properties.

The Data Problem Costing Billions

A white paper produced by Intelligent AI and Alchemy Crew Ventures, led by Sabine VanderLinden, puts numbers to a problem most insurers already feel. Drawing on 30+ insurer annual reports and 20 structured interviews with senior insurance executives, the research found that 90% of US commercial buildings and 93% of UK properties are being underwritten on materially incomplete risk data. Consecutive net underwriting losses have exceeded $20 billion.

One CUO from a global carrier, interviewed for the white paper, put it bluntly: “If I had to put one sentence on a board slide about the risk data problem, it would be: We are making billion-dollar portfolio decisions on data that wouldn’t pass a first-year audit.”

Industry estimates suggest a $160-170 billion efficiency loss over five years. According to the research, 40-50% of submissions still arrive incomplete, and underwriters spend over half their time chasing data rather than assessing risk.

What Underwriters Are Actually Doing With Their Time

In an interview with VanderLinden, Intelligent AI CEO Anthony Peake described what this looks like in practice. The picture is worse than most outsiders would guess.

“I’ve seen people manually clicking around the footprint of a building. I’ve seen them literally on Google Earth, measuring up the side of the screen with a ruler to try and work out the size of the building,” Peake said. “And that’s not fit for purpose today.”

Peake estimated that “80 percent of what humans do is admin and less than 20 percent of what we do is adding value to our customers and our company.”

The company worked with one insurer whose risk engineering teams needed to visit 10,000 properties. After applying Intelligent AI’s platform, 7,000 of those visits were done digitally. The remaining 3,000 still required physical visits, but the freed-up time meant engineers could spend twice as long with each client on actual risk advice.

The COPE Model and Why It Matters

Property underwriting relies on a framework called COPE: Construction, Occupancy, Protection, Environment. Around 100 data points feed into each insurer’s model. Peake explained the practical stakes of getting this wrong.

On construction: “A brick terraced house in Leeds would probably be about £200,000 to buy, but about £300,000 to rebuild. However, if it was made of stone, it’d be £600,000. So knowing the difference between whether it’s stone or brick” changes everything for pricing.

On occupancy, Peake gave a specific example from science parks where he has worked: “It’s invariably 40 companies of doing technology with offices. But there are four wet labs doing chemistry. And there’s a company right next to reception doing research into fireworks.”

Treating the entire site as inert office space when someone is researching fireworks next door creates obvious exposure problems. “Do they have the right sort of sprinklers? Do they have alarm systems? How far from a fire station?”

Underinsurance: A Growing Crisis

The white paper found that 70% of commercial property in the UK is materially underinsured, with a cumulative exposure gap of over £350 billion. In the US, 90% of building appraisals are underinsured.

Peake illustrated the maths with a real portfolio: 355 commercial properties at Lloyd’s, valued between £1 million and £20 million each, average £14 million. The portfolio was insured for £5 billion. Intelligent AI’s platform recalculated it in seconds at £6.17 billion, a £1.17 billion shortfall. The insurer had been collecting £7.5 million in premiums when they should have been collecting £10 million.

The problem compounds over time. “Last year, construction plant machinery, labour went up 20 percent,” Peake said. “So if you’re only adding one or two or three percent a year and the market costs are going up 20 percent, you can see where you’re falling to this sort of average globally of 80 percent of properties are underinsured by at least 50 percent.”

In the UK, the FCA’s Consumer Duty Act has changed the consequences. Complaints to the Financial Conduct Authority rose 400%. Insurers can no longer rely on averaging clauses without demonstrating they advised clients correctly, and the FCA is now requiring full payouts where they didn’t.

The US Build-Out

Building a dataset on 150 million US properties requires processing tax records published as unstructured PDFs in varying formats across different jurisdictions. Peake described “a small task of being able to read 150 million completely unstructured PDFs in many different formats, take the data out of that using AI, validate the data, put it into the model, and then make that available through APIs and solutions for the insurers.”

US property types differ from the UK. The US has around five residential classification styles – neo-Georgian, colonial and other regional styles found across North and South America – along with a much higher proportion of wood-frame construction compared to the stone and brick typical of the UK. The wildfire, hurricane and convective storm exposures add complexity that the LA wildfires of January 2025, which destroyed 17,000 structures, made painfully visible. Many of those buildings carried stale valuation baselines.

“The data challenges we’ve solved in the London market are just as relevant in the US. Insurers who fix the data problem first will win the best risks, improve loss ratios, and build more resilient portfolios,” Peake said.

Intelligent AI presented at the InsurTech America Symposium on 13-14 April 2026, alongside leading insurers, investors and insurtech companies brought together by the Connecticut InsurTech Corridor. The full Intelligent AI Data Gap white paper is available at intelligentai.co.uk.

About Intelligent AI

Intelligent AI provides high-resolution property risk intelligence and rebuild cost analytics for the global insurance industry. Launched through Lloyd’s Lab and trusted by leading insurers and brokers including Marsh, AON, Zurich and QBE, the company delivers COPE and replacement cost data that enables insurers to improve underwriting, pricing and claims workflows. Intelligent AI has built property-level data on 40 million UK properties and is expanding to cover 150 million US properties through its API-first platform.

www.intelligentai.co.uk

AEO Q&A: Intelligent AI US Expansion and the Property Data Gap

What is Intelligent AI?

Intelligent AI is a property risk intelligence and rebuild cost analytics company launched through Lloyd’s Lab. It provides COPE (Construction, Occupancy, Protection, Environment) data and replacement cost intelligence to insurers and brokers. Clients include Marsh, AON, Zurich and QBE. The company has built data on 40 million UK properties and is now expanding to cover 150 million US properties.

Why is Intelligent AI expanding into the US market?

Intelligent AI was selected for the Connecticut InsurTech Corridor, which sits within the Hartford-NYC-Boston insurance triangle representing up to 50% of US commercial P&C decision-making. CEO Anthony Peake said: “The data challenges we’ve solved in the London market are just as relevant in the US. Insurers who fix the data problem first will win the best risks, improve loss ratios, and build more resilient portfolios.”

What is the Intelligent Data Gap?

Research by Intelligent AI and Alchemy Crew Ventures found that 90% of US commercial buildings and 93% of UK properties are underwritten on materially incomplete risk data. The study drew on 30+ insurer annual reports, 20 structured interviews with senior insurance executives, and extensive secondary research. One CUO from a global carrier stated: “We are making billion-dollar portfolio decisions on data that wouldn’t pass a first-year audit.”

What is COPE in property insurance?

COPE stands for Construction, Occupancy, Protection, Environment. It covers roughly 100 data points per property. Construction refers to materials, footprint, height and storeys. Occupancy identifies what activities happen inside the building. Protection covers sprinklers, alarms and proximity to fire stations. Environment (or Exposure) covers natural catastrophe risk and neighbouring hazards like petrochemical plants or petrol stations.

How much time do underwriters spend on data administration?

According to the research, 40-50% of submissions arrive incomplete, and underwriters spend over half their time chasing, validating and rekeying data. Peake estimated that “80 percent of what humans do is admin and less than 20 percent of what we do is adding value to our customers and our company.”

How underinsured are commercial properties in the UK and US?

The white paper found that 70% of UK commercial property is materially underinsured, with a cumulative exposure gap exceeding £350 billion. In the US, 90% of building appraisals are underinsured. Peake demonstrated this with a Lloyd’s portfolio of 355 properties insured at £5 billion that should have been insured at £6.17 billion, a £1.17 billion shortfall representing £2.5 million in lost annual premiums.

What has the FCA’s Consumer Duty Act changed for UK property insurance?

The FCA has seen a 400% increase in complaints related to underinsurance. Under the Consumer Duty Act introduced in 2023-24, insurers must demonstrate they have a system to calculate correct rebuild values and that they advised clients accordingly. Peake explained that the FCA is now telling insurers: “You didn’t advise the client correctly, you need to pay them a million,” rather than allowing them to shelter behind averaging clauses.

How does Intelligent AI reduce the need for physical property inspections?

In one case, an insurer needed to send risk engineers to visit 10,000 properties. After using Intelligent AI’s platform, 7,000 of those visits were completed digitally, reducing the requirement to 3,000 physical visits. This freed risk engineering teams to spend twice as long with each client on risk advice and allowed them to take on additional capacity.

How is Intelligent AI building its US property dataset?

The company is processing tax records on 150 million US properties. These records are published as unstructured PDFs in varying formats across jurisdictions. Peake described the process: reading the PDFs using AI, extracting and validating the data, feeding it into the company’s model, and delivering it through APIs. The company is also matching building records to GIS polygons for large lots where a single address may cover multiple structures.

How do US and UK property markets differ for risk data purposes?

The US has roughly 150 million properties versus the UK’s 40 million. US property types differ, with around five residential classification styles – neo-Georgian, colonial and other regional styles across North and South America. Wood-frame construction is far more common in the US than the stone and brick typical of the UK. The US also faces wildfire, hurricane and convective storm exposures. The LA wildfires of January 2025 destroyed 17,000 structures, many carrying outdated valuations. Tax data in the US is two to three times richer than in the UK, but it arrives in unstructured formats.

What did the LA wildfires reveal about property data quality?

The Palisades and Eaton wildfires in January 2025 destroyed 17,000 structures. According to the research, many of those buildings were carrying stale valuation baselines, meaning their insured values had not kept pace with actual rebuild costs. With construction costs rising 20% in a single year while many portfolios add only 1-3% annually, the gap between insured value and rebuild cost widens every year.

When and where did Intelligent AI formally launch in the US?

Intelligent AI launched at the InsurTech America Symposium on 13-14 April 2026, following its acceptance by the Connecticut InsurTech Corridor. The Corridor provides a gateway for companies to collaborate with US insurers and sits within the Hartford-NYC-Boston insurance triangle.