# AI helps insurers price more of the business they want

Markdown copy of https://realalpha.app/s/ai-helps-insurers-price-more-of-the-business-they-want/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Shyam Sankar](https://realalpha.app/builders/shyam-sankar/) · Shyam Sankar is Chief Technology Officer of Palantir Technologies. · Runs Palantir's technology as CTO, having joined as employee 13 and pioneered the forward-deployed engineering model.
- Host: [Shawn Ryan](https://realalpha.app/people/shawn-ryan/) · Host, The Shawn Ryan Show
- Show: The Shawn Ryan Show
- Episode: [Shyam Sankar - Chief Technology Officer of Palantir: The Future of Warfare | SRS #190](https://realalpha.app/v/DTG393KqOGc/), Apr 10, 2025
- Moment: 2:11:36 to 2:14:19 of the episode (3 min), playing on https://realalpha.app/s/ai-helps-insurers-price-more-of-the-business-they-want/; the episode from there: https://realalpha.app/v/DTG393KqOGc/?t=7896

## Quote

> “AIG just discussed at their investor day earlier this week, how they've used **85 AI agents built in Palantir** to automate their insurance underwriting.”
> — [Shyam Sankar](https://realalpha.app/builders/shyam-sankar/) at [2:12:41](https://realalpha.app/v/DTG393KqOGc/?t=7961) · The Shawn Ryan Show

## Takeaways

1. AIG used 85 Palantir-built AI agents for underwriting.
2. Underwriting time fell from three weeks to one hour.
3. Pricing more requests helps the insurer select better risks.

## About this segment

Shyam Sankar says Palantir AI agents cut underwriting from weeks to an hour and let an insurer assess far more requests. The advantage is choosing better risks, not just saving labor.

From “Shyam Sankar - Chief Technology Officer of Palantir: The Future of Warfare | SRS #190”, Apr 10, 2025. [Watch the full episode](https://realalpha.app/v/DTG393KqOGc/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[2:11:48](https://realalpha.app/v/DTG393KqOGc/?t=7908) What could go wrong with this stuff? I mean, could an adversary potentially hack into our hardware and software as much as they could today you know i um i don't think i don't think it changes the nature of the risks the risks are the same you know we still have to care about cyber vulnerability and and protecting ourselves and the adversaries will have those vulnerabilities as well um i think the biggest risk is really of inaction this is very frustrated around the the constant ai safety conversations because one i think it was grossly overdone we're really far away from, if you understand the technology, like the Skynet fear is wildly overblown and almost like a conspiracy just to slow everything down. But the other thing is, it's really about who's the best at implementing these things, right? You can think about AI supply and demand.

[2:12:37](https://realalpha.app/v/DTG393KqOGc/?t=7957) The supply side is the model. These models, they took billions of dollars to build. They can write poetry, they can do all these things, but who cares? Like what's, where's the economic value or the decision advantage on the battle space? That's the demand side. So who's like leading the world at implementing these things. AIG just discussed at their investor day earlier this week, how they've used 85 AI agents built in Palantir to automate their insurance underwriting. So when you look at a solution like that, you could say, okay, it used to take three weeks, very manual. Now it takes one hour. There's a huge labor savings there, but that's not actually what's valuable about it. What's valuable about it is they used to only be able to price roughly 10% of the customer requests because it was just too much work.

[2:13:20](https://realalpha.app/v/DTG393KqOGc/?t=8000) Now they can price 100%. They can take the best and least risky assets and underwrite that business and leave the rest for their competitors. So you're actually saddling your competitors with risks they probably shouldn't be taking because you have a decision advantage in doing that. And so we need to be the best at implementing AI for economic value, whether that's on the battle space or in our companies.

[2:13:41](https://realalpha.app/v/DTG393KqOGc/?t=8021) And I think there's a lot of promise in doing that right now. If you look at Europe, they're still a little bit more than a little bit. But they're asleep at the wheel at the pace at which they're willing to move, the way they want to think about it, the consensus driving versus you feel it in the American commercial sector. People have rolled up their sleeves. They are getting in. They're experimenting.

[2:13:59](https://realalpha.app/v/DTG393KqOGc/?t=8039) They're iterating. Like, AI is just not something you can think your way through. It's something you have to have a lot of experience and, as a consequence, develop expertise in.