The questions I'm hearing most as teams are exploring AI in underwriting are along the same lines: How do we get real value? How do we ensure compliance? How do we get our underwriters onboard? 

In conversations with underwriting leaders like Jean-Marc Fix, VP of R&D Biometric Risks at Optimum Life Re, and Nichole Myers, Chief Underwriter at Ethos, who have both implemented AI within their teams, a few clear patterns have emerged around what actually works.

To briefly highlight their impact, Ethos found that AI adoption has saved their underwriters 30% of their time reviewing records. Optimum Life Re found that AI reduced the time for a first underwriting review by 39% and reduced the time for a second underwriting review by 69%, resulting in 13% annual cost savings for Optimum Life Re.

So following that discussion, here are some real-world lessons from underwriting teams using AI.

‍

Key Lessons for Successfully Implementing AI in Underwriting

Successfully implementing AI in underwriting takes more than simply finding or purchasing the right tool. It can lead to meaningful improvements in efficiency, quality, and accuracy of medical record reviews, but it’s also a significant shift from how things have traditionally been done. That requires a thoughtful approach to ensure success.

Here are nine key takeaways to help you drive adoption, avoid workflow disruption, and achieve measurable results:

1. Make sure you understand your needs

Before anything else, get clear on what you actually need. There are a lot of tools out there, but not all of them will solve your problem. Start with where you’re losing time or where your process breaks down, not with what a tool claims to do.

‍

2. Know what’s out there

Be aware of the opportunities and stay on top of what’s happening in the market. A lot has already been tried. Some things work, some don’t. Talking to peers, going to events (like AHOU!), and learning from others’ experiences will save you time.

‍

3. Involve people from multiple departments

AI in insurance is no longer just about summarizing medical records. It’s about helping your company make better decisions with more usable data. This discussion should involve underwriting, claims, actuarial and IT teams, to name a few.

‍

4. Minimize disruptions for your underwriters

New tools shouldn’t disrupt your underwriters’ workflows. Focus on how the tool fits into your current process, integrates with existing systems, and aligns with how your team already works. If it doesn’t fit naturally, your team won’t use it and it will create disruption over time.

‍

5. Consider buy vs. build

Building your own AI tools gives you more control over the product, but most teams don’t have the resources for that anymore. Things are changing too fast. Gone are the days of building something and being done with it. Now, ensuring a successful tool requires constantly monitoring, adjusting, and improving.

If you choose to buy a tool instead, you can get instant access to state-of-the art technology. However, it makes finding the right partner even more important. 

(If you want to learn more about the decision to buy, build, or partner, check out this great DigitalOwl blog.) 

‍

6. Find the right vendor

If you decide to go with an external tool, it’s critical to understand who you’re partnering with. You need support and you need trust. The vendor needs to be stable, responsive, and able to handle your volume of medical records. An ideal vendor will:

• Incorporate real underwriting feedback to continuously improve
• Leverage proven, state-of-the-art technology
• Have the scale and responsiveness to properly support your team
• Truly understand insurance and underwriting workflows

‍

7. Support your team through change

Like any big change, your team will need support getting used to a new tool. One way to ease adoption is to create superusers – tech-friendly, open-minded, and respected team members who can learn the tool and help support their teams internally.

It’s also important to give your underwriters time to learn and play with the tool before expecting them to use it daily, and to set clear expectations around how it should be used. The tool does not replace human judgment. Underwriters should still be validating and checking outputs, but they also need to know they won’t be held accountable for any mistakes the AI makes.

All of this helps set the groundwork for successful, positive adoption.

‍

8. Set appropriate guardrails

Be clear about what can and can’t be put into the tool. Set boundaries early. Work with IT and security to make sure everything is compliant and safe, while still giving people access. You need both control and usability.

‍

9. Remember, these tools are strategic partners for underwriters, not replacements

AI tools are there to make underwriters’ jobs easier, taking on manual work and giving them more time for higher-level work. They are not a replacement for underwriters' experience, judgement, or humanity, and shouldn’t be thought of that way.

Instead, these tools are made to act as strategic partners, to help underwriters do their best work by making it easier to connect patterns and better understand how data comes together.

‍

What this means in practice

Across teams, the pattern is clear. AI delivers value when it’s implemented thoughtfully, with the underwriters’ real use cases and needs in mind, and when underwriters are given the proper support needed to learn and trust the tools available to them.

When this is done, the results are significant and measurable, but how teams get there has a huge impact on the overall effectiveness of AI. In practice, success comes down to how well these tools fit into existing processes, how the team is supported through the change, and how that change is communicated to underwriters.

If you’re thinking through how this applies to your organization, I’d be happy to connect. You can reach me at whitney.b@digitalowl.com or schedule some time with me.