$LMND Bear Case Debunked: Why Palantir AI Won't Save Legacy Insurers

There is a bear case that circulates around Lemonade Insurance every now and then. The bear case is: any improvements Lemonade makes with AI will simply be copied by legacy insurers thanks to help from Palantir. In the hands of Palantir, surely a big legacy business with lots of money can do anything Lemonade can?

Let’s explore this idea and see why it isn’t the case.

Legacy Insurance Businesses and Leveraging AI: The Reality

First, we have to understand legacy businesses vs modern tech startups. Tech startups often live by the mantra, move fast and break things. They are lean organizations. Do you want something done? You simply DM that employee directly on Slack.

To get a feel for how modern tech companies operate, here’s a memo from Elon Musk on how employees should communicate with each other. It’s a must read!

This is in stark contrast to how legacy insurance businesses work. These companies have a lot of employees, hierarchy, and processes. There are committees, panels, and discussions before anything can get implemented. Most of the folks in leadership have been there twenty years or more. They know the way things are currently done, and they are comfortable doing it that way. That’s how they like it.

Now imagine you’re a new employee or consultant who is hired by a legacy insurer to help them modernize something. Let’s say you’re tasked to improve customer support. “Everyone hates it,” the legacy management tells you. “Help us fix it, our call center employees have high turnover, customers are frustrated when employees give incorrect advice and keep transferring them to other departments. It’s a headache for everyone!”

You have an idea as a consultant for the legacy businesses. AI is becoming advanced now. Studies have shown AI demonstrates IQ scores ahead of 90% of the population. Why don’t we use AI to help automate customer support? Instead of call center employees providing solutions, we replace them with AI to do it faster with less errors and no wait times!

So what happens?

You first must present it to a committee for evaluation. That committee has the customer service management and IT management team on it to discuss feasibility. Let’s see how they react based on organizational incentives:

• Customer Support Management Reaction: Their manager could lose their job if they don’t have people to manage. They’ve got tens of thousands of support associates. They know how to manage people, not AI. This could cost them and their employees their jobs if everything was automated. Therefore, they tell management: AI is unreliable. People want to talk with people. Let’s not risk it. They lobby against its implementation.

• IT Department Management Reaction: The tech department doesn’t have any AI coding staff. And the manager, who has been here for 20 years, is mostly experienced in traditional coding methods, not how to leverage AI. If he starts hiring an AI management team, they might even prove way more efficient than him. If new hires can do what he does in 1/10th of the time, he could be out of a job! Not only that, they would need to spend months mapping out an AI recruitment plan and catching up on AI itself. They also would have to map out the entire process for customer support. They don’t want to do this. The manager tries to convince upper management not to implement. At best, the IT department informs everyone that it will take significant time to execute this project and face significant hurdles.

Not very positive, is it?

Few if any of the legacy management team are skilled with AI. They are skilled at working with people. They have been used to managing people. They have been promoted within their organization for their skills and leadership as a people manager. They aren’t “AI managers”. If AI replaces the people they manage, there is a conflict of interest. Managers will not actively lobby to replace themselves.

These managers will figure out clever ways to defend their positions. If it ain’t broke, don’t fix it! Plus, they’ve seen trends come and go. Wasn’t everyone talking about how important the “blockchain” was a few years ago? We didn’t end up implementing that, and nobody is asking us to implement that now in insurance. AI surely is just a fad like the blockchain. Something new will come along next year too! If something is serious, we can just buy our way out of it.

Don’t believe me? Watch this video of Ajit Jain, the head of insurance operations at Berkshire Hathaway, speaking at Berkshire’s 2025 Annual Shareholder Meeting about AI:

As Ajit says, AI is “fashionable” now. If it turns out to be a big deal, they’ll be “ready” to pursue it then.

However, they will never be able to pursue it fully, because their system and way of operating is ingrained in these previously mentioned legacy incentive structures. Even if Palantir has a way to make AI work for legacy, it will be limited in implementation and scope, because legacy management will lobby against anything that significantly replaces their subordinates with AI agents.

And as soon as you appoint some young, tech-savvy, highly motivated insurance executive as CEO, there will be a massive power struggle between the CEO and upper/middle management who don’t want to lose their jobs.

History is littered with studies proving this point. Blockbuster, for example, famously failed in the face of Netflix. Netflix asked Blockbuster to buy them for $50 million and was laughed out of the boardroom for being so presumptuous. And yet, today, Blockbuster is bankrupt, and Netflix is worth nearly half a trillion dollars. At the time Blockbuster rejected Netflix, they were fat with cash, and Netflix was cash strapped.

By the time Blockbuster saw they were losing real market share to Netflix, it was too late. They couldn’t keep up with Netflix’s product and innovation velocity and investors were lining up to support Netflix. They no longer needed to be acquired.

New innovative companies have proved time and time again, it’s not the big that eat the small, it’s the fast that eat the slow.

Oh, and, in case you are wondering, Geico currently employs 30,000 associates to help customers.

Meanwhile, take a look at what Lemonade has developed below. It’s a working concept of the AI customer support example used earlier in this article that legacy theoretically “refused to implement”.

In addition: Lemonade already automates 45% of customer support tickets from start to finish. Lemonade also revealed that their net promoter score, a metric that measures customer loyalty and satisfaction, actually increases the more they have AI interact directly with customers.

So will Palantir save legacy from Lemonade? No. While legacy organizations will implement some AI solutions, especially around pricing, they are handicapped by a limited incentive structure which will throttle innovation.

ALSO READ: The Truth About Lemonade's Finances & Cash Position