Edison explains: AI for insurance brokers and platforms – threat or opportunity?

Why does AI in insurance matter for investors?

AI is one of the most powerful forces within the transformative technologies megatrend, reshaping industries from healthcare to financial services. The insurance sector, which is built on data, risk assessment and distribution networks, sits squarely in AI’s crosshairs. In the last month, the stock prices of insurance brokers (S&P Composite 1500 Insurance Brokers Index down 10% ytd) and, to a lesser extent, insurance companies (S&P Insurance Select Industry Index) have shown volatility. In Europe, insurtech platform JDC Group^* is down 13% in the year to date, and has also experienced volatility in the last few months. Concern that agentic AI could become an ‘existential threat’ to brokers’ business models is evident. In this note we focus on the expected impact of AI on the sales and distribution side of the insurance value chain (eg brokers and platforms), instead of AI’s impact on internal insurer processes, such as IT, product management, claim management, underwriting and consumer service.

How should we see agentic AI for insurance?

The Massachusetts Institute of Technology defines AI agents as ‘a new breed of AI systems that are semi- or fully autonomous and thus able to perceive, reason and act on their own’. AI agents differ from chatbots, which answer questions and can solve basic problems, as they can ‘integrate with other software systems to complete tasks independently or with minimal human supervision’.

Applying this to the insurance sector, agentic AI could integrate into the existing software systems of brokers and platforms and make its own decisions about which insurance works best for the consumer and at what price. This would result in the creation of AI brokers.

Such a transformation would probably take place in stages, for example, finding the best-fit car insurance as Insurify aims to do. Insurify’s announcement of the sector’s first ChatGPT insurance comparison app sparked a major sell-off of US insurance broker stocks.

Is Insurify actually an AI broker?

Insurify uses ChatGPT to compare online insurer rates using details of the vehicle, user credit history and driving records, with the last two provided by the consumer. The result is an insurance policy quote that must be confirmed by the insurance company. Insurify is a decision support/comparison tool and does not provide a binding quote, meaning it is not truly agentic AI since it cannot make purchases without human intervention.

However, it is a first step: Insurify is replacing the work of selecting suitable insurance policies. It operates as a comparison tool with some AI features, in that it scrapes the web for quotes. In Europe, insurance comparison sites have existed for about 20 years. For example, in Spain, Tuio is a home insurance comparison app, in Germany, Geld.de (part of JDC Group) offers several insurance categories (home, car etc) and in the Netherlands, Independer is a platform where consumers can buy insurance from a broad selection of insurance providers. Going beyond providing a comparison tool does provide some problems.

What are the barriers to entry in Europe?

There are three key areas that limit what AI can achieve. First, there are legal constraints. For example, in the EU, privacy legislation (GDPR) is more robust compared to the US, where regulation is less stringent and varies across states. These constraints could restrict AI’s use of personal data. There are also general restrictions defined by the EU AI Act, the Insurance Distribution Directive (IDD) and the Markets in Financial Instruments Directive (MiFID III).

Second, there may be local barriers. In Germany for instance, insurance advice can only be provided by registered intermediaries, which have responsibility for providing suitable, individualised and compliant recommendations. Furthermore, German car insurers protect themselves by requiring about 60 data points from prospective insurance takers.

Third, there may be consumer hesitation. Consumers might not be eager to change the way they handle their insurance. This could be because they find new methods complicated or because not every consumer is digitally savvy. The level of digitisation in the US is higher than in Europe, as can be seen through Europe’s slower digital adoption growth rates.

As such it looks like truly agentic insurance distribution will be very difficult to achieve.

What could AI’s impact on the sales distribution side be?

According to a recent presentation from JDC Group, insurance sales channels are now organised so that 38% of the insurance market goes through the insurer’s tied agent networks, 25% through brokers, 12% direct to the insurer, 10% through platforms and 15% through other channels.

Key to being successful with AI in insurance is access to data (client and contract data) and having the infrastructure in place to give sound advice, to provide customer support and have direct access to insurer back-end systems.

This knowledge and access is not available at the individual broker level. As such the broker landscape could very well be disrupted by AI. For platforms this should not be the case, as long as they are able to protect their data access and infrastructure. They will be able to use AI in their own processes to become more efficient and compete with other platforms on these metrics.

Industry research from, among others, CFin Research Center for Financial Services forecasts that by 2035 45% of the market could be channelled through AI. Platforms are expected to keep a constant market share, but brokers, direct at the insurer and insurance tied agent network sales, would lose substantial market share. While the transition could be gradual, by 2035, roughly 62% of the market would be online, compared to the current 29%.

On the customer acquisition side, AI will probably play a large role in the nearer future. After social media and search engines, user intentions will also be signalled by AI search engines and large language models (LLMs) such as ChatGPT and Gemini (Google). AI powered ads already lead to 1.8x higher clickthrough rates and 1.5–2.4x better conversion at JDC unit FMK. AI can be used to pinpoint new homeowners or car buyers who are searching for insurance. Insurers could use AI in combination with data from search engines and algorithms to find potential buyers and then launch targeted campaigns at those groups. Therefore, parties in the sales channel with well-defined AI strategies are well positioned to benefit from the AI era.

What about platforms like JDC Group and Lemonade?

Insurance providers such as US-based Lemonade (a vertical insurer) and Germany-based JDC Group (an insurance platform company) offer full-stack insurance solutions. This means that users can buy a variety of insurance products, such as car, liability, travel, bike, care and home insurance, from a single platform. Lemonade, which has end consumers as clients, is a fully owned insurance stack. This model could be challenging as, although Lemonade will have intimate consumer knowledge, it may not have the best solution for each consumer since it only offers one insurer.

In contrast, JDC Group is largely a platform tool for intermediaries, brokers and agents. It also offers software that can manage insurance portfolios for consumers. JDC Group holds the largest insurance (proprietary) databank in the German insurance industry, with contract data covering all relevant carriers over the past 20 years. JDC Group already leverages its data by using an AI agent for customer queries that can answer complex questions on existing contracts. We expect a diversified platform such as JDC Group would be better positioned in the AI transition compared to a vertical insurer.

Where is the value?

We expect the most effective AI applications for insurance will be in the development of AI brokers, platforms using AI to become more efficient and in customer acquisition by AI ads and intelligence.

The combination of consumer data and insurance data could add value, making insurance platforms such as JDC Group especially well positioned due to intimate knowledge about consumers themselves as well as their insured assets and history, and the amount of data about the insurer, such as costs and coverage details. For traditional brokers, the new market environment might prove to be more challenging. In an increasingly digitised society, AI native brokers may have a better proposition.

Conclusion

We are convinced that AI will have a significant impact on the insurance industry in the next 10 years, but what this will exactly look like is uncertain. There are significant barriers to entry, so we would expect the adoption to be slow certainly in countries with low digital adoption. Nevertheless, we would expect to see the rise of AI insurance brokers, which is positive given the ageing population of insurance brokers in Europe.

The potential for AI in insurance is driven by its data intensity and well-defined processes. We see platforms as the most likely beneficiaries of the AI trend, as they sit on mountains of relevant data, generally have more scale and have developed proprietary tools, including comparison software.

Edison insight

The stocks of insurance companies, especially platforms and brokers, have been hit by the AI trend recently. Concern that AI-based apps, such as US insurance app Insurify or Anthropic’s new AI tools, are a threat to existing business models has caused turmoil. This was partly ignited by a report from US-based Citrini Research, depicting the dystopian effect AI could have on the economy and stock market. In addition, Anthropic’s new Claude Opus 4.6 AI model is capable of writing competent software code, which has affected the software sector. We believe that AI is going to have a substantial impact on the insurance industry. However, what the outcome will look like is still open for debate.

Megatrends: AI, robotics, fintech innovation, transformative technology, society and lifestyle: ageing societies (barriers to digital adoption)

*JDC Group is a client of Edison Investment Research.