From the conference circuit and the trade media to the boardroom and investor calls, there is no shortage of news and announcements about how insurance players are integrating and adopting AI technology within their organizations.
Yet amid all the announcements, acquisitions, and rollouts, less is being said about how AI is actually being used by the people doing the work.
How are the claims adjusters, agents and brokers, service reps, actuaries, underwriters, regulators, and all the other professionals who make up the insurance industry using AI in their day-to-day? Is the investment in AI at the top actually changing the way people are working on the front lines?
As user experience designers, we are less interested in what tools and tech companies are rolling out than in how users experience these tools, and in the systems, processes, and workarounds they leverage on their own accord to be effective in their jobs.
That’s why, over the past several weeks, we’ve been speaking with agents and brokers about how they’re actually using AI in their day-to-day work. We’ve been asking them to talk about where they find value, which processes they are using AI to optimize, and how they feel about it all.
What’s emerging is clear: adoption isn’t being driven from the top down. It’s happening from the ground up. People are experimenting on their own, sharing what works, and gradually integrating AI into their workflows, often with or without company support.
So, how are agents and brokers using AI at work?
1. They are using AI to humanize insurance language for customers.
One of the first places AI is showing up, not surprisingly, is in communication.
Insurance has always been difficult to explain. The way policies are written reflects centuries of compounding complexity and can often sound like gibberish to anyone outside the industry. Agents and brokers recognize this gap and are using AI to bridge it, transforming the terminology buried in dense forms—endorsements, exclusions, limits, etc.—into terms that everyday people can understand.
For someone who's 26, buying a house, and you ask, do you know what a hurricane deductible is? And they're like, excuse me, what? I have to explain it in a way that isn't just like me talking to a fellow broker at a meeting. AI has been really good at helping me just lay it out there for them in a way that is accurate and actually makes sense, without all the jargon.
Paul, Associate Agent, Independent Agency
And while clarity is important, our research participants aren’t about to hand client communication over to AI. Every person we spoke to emphasized that outputs need to be reviewed and shaped to reflect their own expertise and voice.
As Melissa, an account executive at an insurance brokerage, explained, “AI is not perfect. So I triple-check everything. I have a really good relationship with all of my clients, so yes, I want my emails to be accurate and clear and easy to read, but I also want them to sound like me.”
2. They are using AI to review policies and compare coverage.
For many agents and brokers we spoke with, policy review is one of the more time-consuming parts of the job.
They are using AI to work through policies, contracts, and proposals more efficiently — comparing documents, identifying differences, and surfacing key details that would otherwise take hours to track down manually.
As Michael, a commercial producer at a small independent agency that was recently acquired, explained, “We can throw in a couple different quotes or proposals, and it’ll tell us the differences… gives us a full breakdown of what’s covered here versus there, what’s missing, what’s different. It’s not perfect, but it gets you 80% of the way there.”
For many, AI is changing how they approach renewal work and policy comparisons, where the volume and complexity quickly compound.
In the past, I would have handwritten everything. Like, I would have gotten my two screens up and put the expiring policy on one and the new one on the other. I would've handwritten the differences, reviewed them, and then it would've taken me a couple of drafts to type it up on a computer. This would take me hours and hours. I've spent an entire day doing it before. Now it’s done in less than an hour.
Megan, 35, Commercial Account Manager, Independent Agency
The work doesn’t go away, but it is shifting from manually combing through documents to reviewing and validating what AI surfaces.
3. They are using AI to plan and draft communications.
For many agents and brokers, communication isn’t just about writing. It’s about staying on top of who to reach out to, when to follow up, and what to say across dozens, sometimes hundreds, of clients.
Agents and brokers are using AI to bring more structure to that process, helping them plan outreach, set reminders, and map communication over time, rather than relying on memory, spreadsheets, or fragmented systems.
I built a skill in Claude to create a whole annual communications plan for a client. We fed it all the benefits flyers, everything relevant to this particular group, and asked it to create an email comms plan. And it spit out the HTML for monthly email messages for the client. There is one for Mental Health Awareness Month, another that reminds them about the preventive services covered by their plan—that kind of thing. Then it creates tasks in ClickUp with the monthly messages to send and assigns them to the account manager.
Melissa, 37, Account Executive, Insurance Brokerage
4. They are using AI to get answers in real time.
AI is also becoming something agents use in the moment, not just to get answers, but to show up better for their clients. Instead of stepping away to look something up, they can ask a question, keep the conversation moving, and explain things in a way that actually builds trust.
This is the easiest way to get low-stakes answers without making yourself sound like an idiot. I can be on the phone with a client, realize they know more than I do on something, and just get an answer right back.
Mike, 45, Personal Lines Agent, Independent Agency
This is especially useful when explaining unfamiliar or complex concepts. Jeffrey, a senior account manager at an independent agency explained how AI helps him quickly reframe complex policy details in real time for customers:
“I recently had a client ask me about a coverage on a certificate—what the advantage of it was. It was a waiver of subrogation. I knew what it did, but it’s hard to explain to someone outside the insurance industry. So I went into ChatGPT and asked, ‘How do I explain this to a client?’ It gave me four sentences. My client was immediately like, ‘Okay, I get this.’ So it’s not just about getting answers for me, but it’s really helping me build a better relationship and trust with my client.”
5. They are using AI to upskill, educate, and flatten the learning curve.
AI is changing how quickly people can get up to speed. For newer agents, it serves as an on-demand mentor, helping them learn terminology, understand coverage, and apply it in real situations without constantly needing to stop and ask someone else.
That’s especially important as experienced professionals retire and take decades of knowledge with them. As one partner and producer at an independent agency that has just invested in an insuretech AI platform called Cara put it, “there’s so much knowledge leaving our industry with people retiring—we need tools like this. Could people rely on it too much? Sure. But to me, it’s another tool in the toolbox. It’s not replacing learning, it’s enhancing it.”
In some cases, that knowledge is also being captured and shared more directly.
There’s so much knowledge leaving our industry with people retiring—we need tools like this. Could people rely on it too much? Sure. But to me, it’s another tool in the toolbox. It actually helps people understand more, not less.
Eric, 41, Partner and Producer, Independent Agency
The work of learning isn’t going away, but it is changing. Instead of starting from scratch, agents and brokers are learning by reviewing, refining, and applying what AI surfaces. And, for many of the research participants we spoke with, checking in with their more experienced colleagues. Several brokers and agents discussed how they’re actively involved in reviewing outputs, validating decisions, and using AI as a teaching tool alongside newer team members.
“We’ll run it through AI, and then sit down and go over it together—what it got right, what it missed, what we’d do differently. It’s not replacing mentorship, it’s facilitating it,” Linda, a 30 year insurance insurance veteran and account executive, explained.
A wild west, full of fences
For all the progress, AI in insurance still feels early. Agents and brokers are experimenting in real time, trying different tools, building their own workflows, and figuring out what actually works in practice.
But this isn’t happening within a clear or consistent framework. In many cases, there’s limited guidance and uneven support, leaving people to figure it out on their own.
Where the fences start to go up
There are also real constraints around how these tools can be used, which, in many cases, are necessary, with carrier clauses restricting agents from uploading documents in ChatGPT. Those restrictions make sense from a privacy and compliance standpoint. But they also create friction, especially when internal tools aren’t as capable or flexible as externally available tools. As one research participant put it, “Internal AI feels like ChatGPT with guardrails… I’d rather just use the real thing.”
And even when the tools exist, getting value from them isn’t always straightforward. Many of the systems people rely on every day still don’t connect cleanly with AI tools, making it harder to embed them into real workflows. In some cases, that means the potential is clear, but the execution isn’t there yet. One agent described wanting to build an AI assistant using their company’s internal tools, but couldn’t because it wasn’t yet integrated with Salesforce.
So while the technology is advancing quickly, the day-to-day reality is more fragmented. People are navigating tradeoffs between what’s possible, what’s allowed, and what actually fits into how they work.
And when the stakes are high, human judgment still takes priority: One participant captured this sentiment best: “If it’s something that’s a big deal for a client… I want to talk to a person.”
As AI becomes more capable, it raises a bigger question: how much of this work should be automated, and how much can be, given the systems, constraints, and expectations the industry operates within?
___
This is just a glimpse of what we’re seeing so far. There’s much more to unpack, especially around trust, workflow challenges, and how roles may evolve as AI becomes more embedded in the industry. We’re continuing to explore those questions and sharing what we’re learning as we go. If that’s of interest, you can follow along here.
Related
by Emily Smith Cardineau
