The specialty insurance market is steeped in tradition that has influenced its working ways ever since those early days. It’s this history that has made London one of the world’s best and most experienced markets for specialty risks.

But the digital revolution and the rapid take-up of AI have the potential to re-shape many industries. The insurance markets are no exception; these technologies will have a profound impact upon the way that risks are underwritten.

The creation of Ki — a digital underwriting platform for the Lloyd’s market — was first announced in May 2020. The aim was to significantly reduce the amount of time it takes for brokers to place their follow capacity, after arranging lead terms for specialty risks. The solution: building an algorithm, developed with support from University College London (UCL), to underwrite risks and automatically quote for business through an always-available digital platform, built with Google Cloud.

Brokers log in to the Ki platform, fill out some details about the risk and can instantly generate a follow line, in as little as 10 seconds. The algorithm behind the platform underwrites the risk and determines how much capacity to offer to that risk.

To deliver this innovation, a first for the London market, Ki raised USD500M of committed capital from two backers, Blackstone and Fairfax.

Underwriting risk in less than 10 seconds

Ki, at its very heart, is an underwriting business. But its unique feature is the way that the algorithm works together with human underwriting expertise to create one of the market’s fastest, most consistent underwriting processes. A quote can be offered to a broker within around 10 seconds of them inputting the risk information into Ki’s platform — a far cry from the two- to three-week lead time common in the market

Ki’s algorithmic underwriting strategy is based on two core concepts: it is data-driven, and technology enabled. Specialty insurance is inherently complex, with subtle nuances. To accommodate this, Ki’s algorithmic strategy sits alongside its wider underwriting strategy.
In 2023, Ki wrote USD877M in gross written premium, delivered a combined ratio of 89.7% (a nearly 7-point improvement on 2022), and USD101M of profit. This strong performance, in only its third year of trading, validates Ki’s approach to digital underwriting.

Why is algorithmic underwriting good for insurance?

Algorithmic underwriting is beneficial for the insurer, the client, and those who artfully curate coverage — the brokers.

For insurers:

Data-driven approaches are consistent and focus solely on the relevant information to evaluate the risk. They are not swayed by personal observation, preference, or bias. And the offer of follow capacity can be as bullish or conservative as the data tells you to be.

As a software system, Ki’s underwriting algorithm can evaluate multiple data points simultaneously for a fast, robust, and accurate underwriting assessment. The algorithm can also be pressure-tested against a variety of conditions, such as natural catastrophes and market cycles. This is much harder to do using human underwriters.

Auditing and optimizing are much easier with an algorithm and digital approach, helping to refine and improve continuously.

Scalability is another huge benefit. Ki’s model is more flexible due to its end-to-end automation made possible by algorithms, allowing it to adjust smoothly to both soft or hard market conditions.
And importantly, Ki’s underwriting algorithm can also incorporate a diverse set of data points from within and outside the business, ranging from portfolio data at the point of quote to property characteristics derived from satellite data on individual properties. This allows Ki’s underwriting to be both accurate and efficient.

For brokers and clients:

Because of the accuracy and continuous optimization of the algorithm, Ki can offer increased line sizes and open up capacity across the portfolio in less risky geographies or classes to offset higher-risk policies.

The firm prides itself on being easy to work with. The platform is available to brokers 24/7 and ready to quote consistent and reliable follow capacity from four Lloyd’s of London syndicates — Aspen, Beazley, Ki 1618, and Travelers.

Brokers are offered follow capacity for eligible classes every time, as long as one condition is met: a “nominated syndicate” is on the slip. A nominated syndicate is a Lloyd’s syndicate that Ki recognizes as a leader in its respective class of underwriting.

For brokers and clients alike, certainty of renewal is attractive, particularly when facing challenging market conditions. And if you compound that with increased capacity being available when policies come to renew, in most cases, Ki is a sustainable partner for a long-term mutually beneficial relationship.

Ki brings significant efficiencies to brokers with its fast response times by providing them robust follow capacity underwriting with minimal involvement on their part, so they can focus on the lead placement and negotiating (coverage, cost) on behalf of their clients.

The technology

Using data to understand risks isn’t new, but what is new is the access to cutting-edge data science and machine learning. And at Ki, a healthy dose of specialty market expertise is added to create something truly unique that is tailored to the market dynamics at any given time.

From its very beginnings, when Ki collaborated with leading academic talent from UCL, the company’s proposition has been powered by cutting-edge algorithmic science. The Ki algorithm of today is enabled by more than 50 Machine Learning (ML) models and maintained and iterated in-house by a multidisciplinary team. And as mentioned above, it can draw in data from a wide range of internal and external sources.

Ki also uses Large Language Models (LLMs) to help structure the vast quantities of data locked in policy and claims documents. This is now being leveraged to provide richer data sets for its ML models.

Ki, and its diverse and talented teams, are committed to continuous innovation and often contribute to academic research, for example, by recently publishing research on simulating the specialty insurance market.

The future of algorithmic underwriting

There’s a huge range of opportunities for algorithmic underwriting, from small efficiencies across the value chain to further disrupt traditional processes and enhance the human expertise that will still be important to have. As algorithmic underwriting gets more sophisticated, it will allow an insurer to holistically manage a portfolio across classes at the point of quote. This is difficult for human led syndicates to deliver because of the disparate nature of their data sets, people, and processes. Making better individual risk selections will have a positive impact on the loss ratio, resulting in a portfolio that’s well balanced and profitable.

Utilizing LLMs to tap into the vast pool of unstructured data available in insurance will provide a huge informational boost to algorithmic underwriting and portfolio management. LLMs will also replace notoriously time consuming, repetitive tasks such as wording reviews and will allow process-heavy work to be streamlined, making it more accurate and more efficient. Ultimately, all of this will reduce expenses and improve an insurer’s combined ratio.

The Ki difference

While algorithmic underwriting underpins Ki’s proposition, the company balances and complements this with an expert understanding of the market. Ki’s underwriting approach is powered by technology, but it’s overseen and perfected by people who are specialists in insurance, who bring an understanding of events not in the data’ and of the real-time market dynamic shifts that occur. Ki’s teams are a blend of technology and data talent, working collaboratively with more traditional insurance roles such as class specific underwriters, account management, and business development to build relationships in the market.

It’s a unique approach in the market with a truly scalable proposition rooted in underwriting discipline and a digital-first environment.

This article was created for Gallagher Re’s Global InsurTech Report, August 2024.