There's a thing I keep seeing that I can't stop thinking about.
Somebody opens Claude, drops in their bank statement, and says "help me build a budget." Claude reads through it, sees the line item for Allstate or Geico or State Farm, and says: wait a minute. I know the cars you have, I know the price, I know your geographic location. That's a little high for what I know. Step one - let's go shopping for auto insurance.
There’s a new shopping cycle in the age of AI. People are no longer typing "auto insurance quote" into a search bar when they’re in-market. AI just noticed the opportunity based on their conversation and goals, and now it's shopping for insurance on the consumer's behalf.
I had a great convo with Michael and Courtney Weaver about this on The Insurance Buzz podcast recently. Here’s more of what we covered.
The first version of using AI is opening up ChatGPT and using it like a Google replacement. You ask it stuff. It answers. You start to notice, oh cool, it remembers me, its predictions about me are pretty good. A lot of people stop there.
Operating with AI is different. Every day I come in and ask about the problems I'm trying to solve in the business. Small ones, big ones. Three or four years ago I'd solve them with code. Now I solve them with AI. I've got a lot of agentic agents running in parallel - looking at my calendar, prepping me for meetings like this podcast, doing research on the guys I'm about to sit down with.
It's muscle memory. Kind of like going to the gym. Today I'm going to get a little better at running my business, a little more efficient, because I actually solved a real problem with AI. When you make that leap, you're in new territory.
Keeping this simple.
When you open ChatGPT or Claude, it's turn-based. You ask a question, it gives a response. You can get good at prompting, you can steer it, but you're sitting there interacting in a two-way loop.
Agentic AI uses logic and reasoning to decide what to do next. It can connect to your calendar or an external system, do things start-to-finish, and ask itself how it's doing along the way. Instead of the back-and-forth loop, you've got an autonomous employee running in the background.
For an insurance agency, most of the value shows up at the top of the funnel. If you're growing your book through volume - buying third-party leads or lucky enough to produce a lot of first-party - there's a mountain of work between "we're going to start acquiring leads" and "these are bindable folks."
Figuring out who's interested. Who's qualified. Who hasn't had insurance in six months. Who's not good for captive. Who's actually looking for an SR-22. Who's a non-standard lead.
At scale, that work is pretty systematic. Ask a couple of questions, bucket the person into a scenario. Spending 40 minutes on a phone call to arrive at "not a fit" isn't the best use of a licensed producer's time. It's not the best use of the consumer's time either.
Humans are best reserved for the nuance. The actual conversation. The decision, the logic, the licensed activity that only a person can do. If we make producers as efficient as possible so they only spend time on that, the volume you can drive is astonishing.
Back to the bank statement scenario, because I don't think most agents have the full picture of what it means.
The more you use AI, the more it stores about you. That memory - the context window - is the underlying magic. Now imagine your Claude understands your vehicles, your home, that your family is growing, that your grandma is moving in. It knows the moments in your life when shopping insurance actually makes sense.
Instead of you filling out a new lead form on EverQuote from scratch, Claude connects that memory to a world of possibilities and sort-orders the agencies best for you. Then a human producer plugs in on the other end and says: I see your file, I understand your scenario, let's talk through the ad hoc stuff.
This is a brand-new kind of ad. That's the opportunity we're playing for. And you don't have to be that old to remember the early days of the internet - there's an outsized advantage to being an early mover.
Here's the surprising part. Consumers and carriers are adapting to this technology faster than the agencies are. It's genuinely surprising to me, because agents know where they need to go. They see it. But there are team members literally saying, "I'm not using that."
Michael's take on this was blunter than mine, and honestly, fair. Consumers already use GPT every day. They see the value. Their expectation going into shopping for financial services is already there. If your producer opens with "let's start with your vehicles" and the consumer's already living six steps ahead of that, you've lost the interaction before it started.
Change always creates friction. There are fair arguments on all sides. But if the value is greater than the cost, it always pushes through. That's what's happening.
Michael asked what I'm a champion of. I said: just being unique.
There's too much of everyone doing the same thing right now. Whatever your thing is - a niche, a category, a way of talking to your customers - go deep. Go deeper than you think. I promise there's more there, and fewer people have gone that far, so the value compounds.
Courtney called it your freak flag. That's it. It's getting noisy out there. Unless you've got yours flying, how are people going to know?
Thanks to Michael and Courtney for having me. Full episode below.
