Why were Ajit Jain’s comments about AI in insurance notable?

Ajit Jain’s comments were notable because he suggested that AI is still years away from handling insurance decisions like pricing policies or settling claims. That stood out because AI-first insurers like Lemonade have already used automation and machine learning in parts of the claims and underwriting process for years.

How is Lemonade using AI in insurance?

Lemonade uses AI and automation to handle parts of the insurance process, including customer onboarding, policy management, fraud detection, and claims handling. Some straightforward claims can be reviewed and paid automatically, while more complex or uncertain cases can still be routed to humans.

Why does AI matter for renters insurance?

AI matters for renters insurance because premiums are often small, which makes human-heavy claims handling expensive relative to the policy value. Automation can help reduce overhead, speed up claims, and make low-dollar insurance products more economically viable.

Does AI replace every insurance decision?

No. AI does not need to replace every insurance decision to be valuable. The more realistic model is that AI handles routine, high-confidence tasks quickly, while complex, unusual, or uncertain cases are escalated to human review.

What is the risk for legacy insurers?

The risk for legacy insurers is that AI-native companies may improve faster, operate with lower overhead, and deliver a better customer experience. Legacy insurers may still have scale and capital advantages, but older systems and slower workflows could become a disadvantage as AI capabilities improve.

Berkshire Just Told on Itself

Rebellionaire Staff
2 hours ago
4 min read

https://www.youtube.com/watch?v=C7DW5m89olo

There are moments when a legacy company doesn’t just sound cautious.

It sounds late.

That was the vibe coming out of Berkshire Hathaway’s annual meeting when Ajit Jain, Berkshire’s legendary insurance executive, was asked about AI in insurance. His basic answer? AI may help with repetitive work, but when it comes to things like pricing policies or settling claims, he’s skeptical. That kind of judgment, in his view, is still years away. [1]

That would be a reasonable answer in 2014.

In 2026, it lands a little differently.

Because the awkward part is this: Lemonade has already been doing pieces of this for years.

Why was Ajit Jain’s AI answer such a red flag?

The red flag isn’t that Jain is cautious. Insurance should be cautious. Bad underwriting is how companies blow themselves up slowly, then all at once.

The problem is that he described AI-powered pricing and claims settlement like it’s some far-off science project.

It isn’t.

Back in 2016, Lemonade’s claims bot, AI Jim, reportedly paid a claim in three seconds. The system reviewed the claim, checked it against the policy, ran anti-fraud algorithms, approved the payment, sent wiring instructions, and closed the claim. Three seconds. Zero paperwork. [2]

So when a Berkshire executive says this kind of thing is still years away, the natural reaction is…

Years away from whom?

Because for AI-first insurers, this isn’t some dreamy “future of work” conference panel. It’s infrastructure. It’s operating leverage. It’s the whole point.

What is Lemonade doing that legacy insurers aren’t?

Lemonade was built around bots, machine learning, and automated workflows from the beginning. That matters because insurance is basically a giant data problem wearing a boring suit.

Who is likely to file a claim?

How should the policy be priced?

Is this claim legitimate?

Can this be approved instantly?

Does this need human review?

Legacy insurers have historically answered those questions with layers of people, departments, forms, call centers, adjusters, and old systems duct-taped together behind the scenes.

Lemonade’s model tries to route as much of that as possible through software first. Not because software is cute. Because humans are expensive, slow, inconsistent, and — in low-dollar insurance lines — sometimes economically impossible to justify.

Renters insurance is the easiest example. If someone pays $60 a year in premium, you can’t have a person spend an hour reviewing a small claim and expect the math to work. The overhead eats the product alive.

That’s where automation starts to matter. A lot.

Why does “AI-only” change the conversation?

Daniel Schreiber, Lemonade’s cofounder, recently wrote that “AI-first” may soon sound quaint. His argument is that companies won’t just bolt AI onto old workflows. The stronger version is to redesign the company around AI from the ground up. [3]

That’s the part legacy insurance should be sweating.

The early AI use case is simple: take a human task and make it faster.

The next phase is different. You stop designing horse-shaped machines.

That analogy matters. Early engines replaced horses one-for-one. Then engines became so powerful that no one asked, “How do we build a better horse?” They built cars, planes, factories, ships, and machines that made the old comparison look silly.

AI in insurance may follow the same pattern.

At first, it helps a human adjuster.

Then it routes simple claims.

Then it prices risk more dynamically.

Then it starts identifying patterns humans wouldn’t notice because the system is watching millions of interactions across the platform.

Eventually, the question changes from “Can AI do what our people do?” to “Why did we design this company around people doing repetitive risk math in the first place?”

That’s uncomfortable.

Also probably true.

Is this really a Berkshire problem?

To be fair, Ajit Jain is not some random executive. He has been one of the great insurance minds of the last several decades. Berkshire’s insurance machine became a monster in part because people like Jain understood risk better than almost anyone else.

So this isn’t a “smart guy dumb” story.

It’s more interesting than that.

This is about what happens when a world-class operator in the old system looks at the new system and sees a tool, not a rewrite.

That’s the trap.

The best fighter pilot in the world is still limited by the human body. Remove the pilot, and suddenly the aircraft can be designed differently. More G-force. Less life support. Different shape. Different mission.

Same idea here.

A brilliant human underwriter may still be brilliant. But if the future is real-time data, instant claims routing, dynamic pricing, automated fraud detection, and AI-native customer interactions, the legacy structure starts to look heavy.

Not useless.

Heavy.

Why should investors care?

Investors should care because disruption usually looks dumb before it looks obvious.

At first, the incumbent says the new thing is too small.

Then it’s too niche.

Then it only works in simple categories.

Then it won’t work in “serious” lines of business.

Then the cost structure starts to break.

Lemonade still has plenty to prove. It’s not Berkshire. It’s not printing endless insurance profits. It has to show that the AI-first model can scale across bigger, messier categories like home and car without underwriting mistakes blowing through the income statement.

But dismissing the AI piece as years away misses the point.

The question is not whether AI can replace every insurance decision tomorrow morning.

The question is whether AI-native insurers can keep improving faster than legacy insurers can retrofit old systems.

That’s a very different question.

And frankly, it’s the one Berkshire should be asking.

Editor’s note for investors

The real story here isn’t “Lemonade good, Berkshire bad.”

Too simple.

The better takeaway is that AI disruption doesn’t always announce itself with a giant, cinematic moment. Sometimes it shows up as a three-second claim. A lower expense ratio. A better customer experience. A pricing model that learns faster. A company that doesn’t need to drag a century of legacy process into every new product.

Berkshire may still be Berkshire.

But when one of the most respected insurance operators in the world talks about AI like the train hasn’t left the station yet, it’s fair to wonder who’s actually watching the tracks.

Because from where we sit?

The train is moving.

Resources

[1] Barron’s. “First Real Question Is on AI and Its Impact on the Insurance Business.” Barron’s, May 2026. (Barron's)

[2] Lemonade. “Lemonade Sets a New World Record.” Lemonade Blog, Jan. 2017. (Lemonade)

[3] Schreiber, Daniel. “After ‘AI-First’ Comes ‘AI-Only.’” Substack, Apr. 30, 2026. (dschreiber.substack.com)

[4] Lemonade. “Lemonade Announces First Quarter 2026 Financial Results.” Lemonade Investor Relations, Apr. 2026. (Lemonade)