Is commercial insurance pricing ready for AI?

In this post, we share key learnings from AI Forum Panel Session 1: Using AI to enhance pricing.

Last month, we proudly hosted our biggest event to date — the inaugural AI_Forum. Produced in collaboration with our knowledge partners, Boston Consulting Group, our ambition was to strip away the hype and discuss the real applications of AI in commercial insurance. We’ll be posting weekly to share the insights from each session.

In this post, we’ll cover Panel Session 1: Using AI to enhance pricing. Watch the full-length video and read our summary of the session below.


  • Richard Hartley, CEO, Cytora
  • Siraj Khaliq, Partner at Atomico 
  • Will Hawkins, Head of European Research, KBW 
  • Mike Hood, Chief Business Actuary, AXA XL  

Is now the right time to apply AI to commercial insurance pricing?

The consensus of the panel was ‘yes’; now is the time to apply AI to insurance pricing for mid-market and SME segments where risks are homogenous and losses occur frequently. 

Assessing large and complex risks where data is sparse requires human judgement and decision making, but AI can still augment and streamline the parts of the pricing process.

Mike Hood, Chief Business Actuary at AXA XL, mentioned that in the large commercial space, an underwriting decision is the result of many small decisions. AI can help to streamline these decisions to enable a more accurate and timely underwriting workflow.

How is AI enhancing insurance pricing today?

The panel discussed how AI can provide enhanced insight into expected claims activity. Ultimately this enables more risk-reflective pricing and helps insurers to provide the right level of protection at a fairer price that reflects the true level of exposure.

AI helps insurers to set a more accurate premium for a risk, based on better external data, and helps to bridge the gap between the expected loss cost and the actual loss cost. This is an area we are working in with a customer who cares very much about risk premium adequacy.

Richard Hartley, CEO at Cytora,

This NOT a revolution…

Will Hawkins, Head of European Research at KBW, highlighted that the impact of AI on insurance pricing is evolutionary rather than revolutionary. When applied to insurance pricing, AI doesn’t completely change the existing business model, it makes the existing business model more efficient and makes insurance more available to a wider market.

Is black box underwriting a problem or a distraction?

The panel agreed that black box underwriting is a relevant issue as model explainability is essential in commercial underwriting. Insurers want to understand what’s driving the outputs of a model so that they can justify why they are charging a certain price to brokers and customers.

This is a reality rather than a problem. We view it as a product area. Cytora produces key drivers to help explain why our model is delivering a specified score. A key driver could be the distance of a building from a fire station or the financial health of a company.

Richard Hartley, CEO at Cytora

Actuaries vs Data Scientists

The panel agreed that Actuaries will not be replaced by Data Scientists. The role of the actuary will evolve, but actuaries are invaluable as they understand the nuances of insurance. In some cases, it can take years for an AI model to outperform a really good rules-based model.

Advice for those starting their journey with AI…

1. Projects must be driven by senior management.

The industry has already changed massively. For a large transformation project, it is critical to have sponsorship from a named member of senior management. There are so many moving parts that it’s difficult to get things done. You need someone driving it who has real decision-making authority, otherwise, you will not get very far.

Will Hawkins, Head of European Research, KBW

2. Build a data advantage.

Focus on assembling a data advantage. Insurance delivers its value proposition through prediction. Within the next ten years, the ratio of data to uncertainty will increase exponentially, and those who can harness that data will win at prediction.

Richard Hartley, CEO, Cytora

3. Partner with a company that really understand the technology.

Pay attention to data, don’t throw things away. Partner with a company that really understands the technology and take advantage of that expertise. Partnerships are important.

Siraj Khaliq, Partner at Atomico

For more information about AI Forum please visit If you’d like updates to be sent directly to your inbox, please subscribe to our newsletter below.