In 2009, Aaron Kassover founded AgentMethods to give independent insurance agents the kind of website and marketing platform that only big carriers could otherwise afford. Fifteen years later, it’s the platform behind thousands of senior market agents. But building AgentMethods also gave him a front row seat to a different problem: agents who had built loyal books of business but had stopped taking new customers entirely—not because the leads had dried up, but because they no longer had the hours in the day.
That observation is what led to SureBook, an AI back office built directly into an agent’s calls, CRM, and inbox. I asked Aaron about what 15 years serving the same customer taught him, why SureBook had to be built around AI from day one instead of bolted on, and what he wants people to understand about where this is headed.
You’ve spent over 15 years building AgentMethods for independent insurance agents. What did that decade and a half teach you about this specific customer that most outsiders would never guess?
It’s easy for those of us who live and breathe software all day to forget that our perspective on technology isn’t a universal truth. Here’s a simple example: when buying software, would you rather take a free trial or set up a sales call? Almost everyone who builds software gives the same answer, free trial. We learn by getting our hands on things, poking at them, testing them. So when I launched AgentMethods, we started by selling the way I wanted to buy: with a free trial.
Insurance agents are people people. They’re good at their job because they’re good at talking with people, understanding people, asking the right questions. So it only makes sense they’d prefer to use what they’re good at when making a purchase. They don’t want a trial, they want to talk to someone.
A few years in, we ran an experiment. Next to the “start your free trial” button, we added an option to “schedule a demo.” Over the next 12 months, we signed up hundreds of agents via a demo, and I can count on one hand the number who became customers after a free trial. That same pattern shows up in virtually every product decision: the layout of a screen, how you label things, all of it has to map back to your customer. It’s incredibly easy to make thousands of design decisions around your own taste, and unless you are the target customer, that usually leads to a worse product. So what did a decade and a half teach me? Something simple and cliche: know your customer.
SureBook goes much deeper into an agent’s daily workflow than a website platform ever could—into their calls, CRM, and inbox. What was the moment you realized the next problem worth solving wasn’t marketing anymore, it was the back office?
Marketing is still a huge problem to solve. But if you look at the growth trajectory of an insurance agent’s business, there’s a shift at some point. What gets a new agent to their first 100 customers will carry them to 500, but unless they change how they approach the business, they get stuck and plateau. Sales and marketing stay important, but the grind of supporting a growing book of business eventually consumes an agent’s day.
I spent time with AgentMethods customers last fall, watching them work, seeing where they got stuck. These were successful agents with great businesses and bigger ambitions they hadn’t met. One comment, from an agent at a hotel bar after a conference, hit hard: “I’ve stopped taking new customers. No more marketing. I just don’t have the time.” I dug in. He had loyal customers and a steady stream of referrals, real opportunities to pursue. He just couldn’t pursue them, because his days were bogged down in scheduling, communications, CRM data entry, and follow-ups. Six or more hours a day, with no time left for growth.
Building a second product for the same customer is a different challenge than building the first. What did you have to unlearn from AgentMethods, and what carried over directly?
The stakes are higher. AgentMethods developed incrementally. Our first release was very simple and looked nothing like the platform today, and we could test an idea quickly and expand over time based on feedback. With SureBook, we’re selling into an established customer base and handling mission-critical operational problems. The bar is significantly higher from day one, with far more integrations to get right. I’ve had to unlearn shipping simple and iterating slowly. We’re still iterating like crazy at SureBook, but the starting bar is much higher.
What carries over is the edge that 15 years gives us. We know far more about how the business operates today than we did at the start, and we have deep relationships with agents who’ve been part of the design process since day one. Those relationships are what let us start the Founding 50 from a place of trust.
SureBook is built around AI from day one rather than bolted onto an existing workflow. What did you learn building AgentMethods that convinced you the old CRM-plus-AI-feature approach wouldn’t be enough?
CRMs are doing a lot of great things with AI now, summarizing client history, drafting emails, letting you query your data in plain language. All useful, but limited to the data inside that one platform. The problems we saw in our customers’ businesses aren’t single-system problems. They’re cross-system, cross-workflow problems that need wide context to solve, and AI inside a CRM can’t reach a problem that lives around the CRM, not inside it.
Software has traditionally been good at black-and-white problems and clearly defined processes. Marketing automation is a good example: you define if/then/else workflows to trigger templates. It’s scalable, but rigid, and it doesn’t allow for the personalization that comes with real human relationships. Modern LLMs change that. We don’t have to force everything into a black-and-white problem to build useful software around it, and that’s exactly what lets SureBook go deeper into a client’s workflow than traditional software ever could. We’re connecting into as many of our customers’ systems as possible and layering in deep knowledge of the products they sell and the compliance rules they follow, so the platform frees them to focus where they add the most value, without needing to hire operational staff.
You’re capping the founding cohort at 50 agents and onboarding them personally. Why that approach instead of scaling fast, and what does it say about how you think about trust with this customer base?
SureBook already works amazingly well. To the agents using it, it’s transformative. There’s a lot we could do, but we need to stay focused on our customers’ actual needs while we’re moving this fast. Limiting the founding cohort to 50 agents lets us give access to the right agents and work closely with them to optimize the platform around real use, and it lets us battle-test the platform properly. By the end of the upcoming enrollment season, we’ll have a deep, evidence-based understanding of SureBook’s impact.
As we expand beyond agents who already know AgentMethods, they’re going to ask reasonable questions. Who are we, is their data safe, does it work as promised, can they trust us? We need real case studies and results to answer that. Ultimately, trust is earned one agent at a time. I’d rather start with a foundation of 50 agents who trust us deeply than 500 strangers.
What do you want people to understand about SureBook right now, at this stage, that they might not get just from looking at the site?
The potential for AI goes so much deeper than chatbots. Agents who put AI at the core of their operations can scale their business, handle more customers, and deliver a level of personalized service a traditional agency can’t match. Getting there takes real work, connecting systems, building in business process and domain knowledge, making sure the AI performs consistently and accurately. That’s the work SureBook is doing. Our vision is to make the next generation of independent insurance agents more relevant than ever. What we have today is already powerful, and we’re just getting started.
If you had to convince a skeptical agent in sixty seconds that SureBook actually changes their day, not just their software, what would you say?
“AI is here. Next Tuesday, when a customer leaves you a voicemail or sends an email you can’t answer right away, you should 100% expect the next place they take that question is ChatGPT.”
While AI can never match the level of care and personal relationship you bring, it gives a very good, and immediate, answer. And next Tuesday, not in some hypothetical future, your clients are going to see that answer and ask themselves: how is my agent better than ChatGPT? The honest answer is that you bring understanding, relationships, and humanity no AI can match. But in some ways, you’re not better than ChatGPT either. The path forward isn’t to fight AI, it’s to enlist it on your side. The agents who win pair the things only a human can do with the things AI does well.
SureBook makes that real. For an agent using it, next Tuesday looks very different. Even with wall-to-wall meetings, the response to that voicemail is drafted and ready to send between meetings. The customer gets help in minutes, no need to ask ChatGPT. Meanwhile, SureBook is working behind the scenes, updating the CRM, tracking open client items, and making sure every opportunity keeps moving forward.
Key Takeaways
Sell the way your customer actually buys, not the way you would. Agents wanted a conversation, not a free trial, and the data backed it up almost unanimously.
Growth plateaus are often an operations problem wearing a marketing costume. An agent who “stopped marketing” hadn’t run out of demand, he’d run out of hours.
A second product for an existing customer starts from a much higher bar than the first. There’s less room to ship simple and iterate slowly when you’re already trusted with mission-critical work.
AI earns its place when the problem spans systems, not when it’s bolted onto one. Point solutions inside a single CRM can’t solve problems that live between systems.
Trust compounds in cohorts, not headcounts. Fifty agents who trust you deeply beat 500 who don’t yet know you.
The real competitive question isn’t AI versus humans, it’s who pairs them best. Clients are already asking AI their questions; the opportunity is making sure their agent answers first.
