Ask an independent agent about technology, and you probably won't hear much resistance. What you will hear is fatigue.
That fatigue, whether technology or AI, matters because the usual story about agency tech is that agencies (and agents) are slow to change. But the data tells a different story. 46% of independent agencies now report using AI, up from 15% in 2024. Another survey found 68% of agents are “somewhat” or “very likely” to increase their AI use in the next 12 months.
That same survey found that only 8% of agencies use AI regularly and strategically. The interest in technology, and AI specifically, is high. Everyday use, however, is not. So where does adoption stall?
Agency leaders cite familiar barriers: not knowing what the tools can do (60%) and security and privacy concerns (48%).
The Big I’s Agents Council for Technology (ACT) report adds a sharper observation. Many agencies, it says, aren't resistant to technology. They've been worn down by constant system changes, retraining, and overlapping solutions. Fatigue shows up as stalled adoption, uneven use, and staff quietly working around approved tools. The Big I study also found that the biggest technology challenge for agencies, as in past years, is dealing with multiple insurance provider interfaces.
Matt Wielbut, chief technology officer at Openly, says the first fix is to “stop asking agents to remember things.” He hears too often that a quote only works after clicking in a certain order, sometimes several times before the system saves. “That should be unacceptable,” he says. “An agent shouldn’t have to remember a Nintendo ‘cheat code’ to get a quote to work.”
A new tool doesn’t arrive alone. It brings a new login, a training session (or several), a new place to key in the same client details, and a new step to remember at 4:45pm on a renewal day.
The average independent agency has 9.9 staff members. That's a small team, and few have dedicated IT departments. Each new tool lands on the same people who are also answering the phone, processing applications, and providing tailored guidance to clients.
The invoice shows only part of the expense. “The hours that go into implementation and training are at least visible,” says Aaron Tekippe, vice president of IT and security at Openly. “The security work is not. Vetting the vendor, locking down access, and adding protective software are all necessary, but in an agency with no dedicated IT team, they’re the first things to slide. Skip them once and it’s a small gap. Skip them across five tools and it’s an accumulated risk nobody is tracking.”
There’s a cost in confidence, too. Joe Harman, vice president of engineering at Openly, puts it as a question: “Do you actually trust that the tool works, or are you double-checking it anyway? Are you confident that it’s helping, or do you now have an additional low-level anxiety about it?”
There is a counterpoint where some tools do pay off. ACT reports that AI is already taking over high-volume, repetitive work in quoting, underwriting, and client service. Vendors forecast 15% to 40% efficiency gains in back-office work, though those are forecasts, not documented results.
A tool fits when it works inside the day an agent already has, whether that’s the agency management system, customer management system, or a rater like IVANS. A tool that fits removes a step rather than adding one. Most importantly, it behaves the same way each time, so nobody needs a workaround.
Tekippe frames fit as two tests. “It has to complete or augment a process you’ve actually defined and can measure, because without that you’ll never know whether it paid off,” he says. “And it has to integrate natively with the systems you already run. A workaround to integrate a tool is just another manual step, and a roadmapped integration is a promise, not a return.”
Harman adds a practical check: will anyone come back to it? “You have to have a compelling reason to come back to it,” he says. “If you don’t need the tool to get your job done, or it doesn’t obviously materially help you, it’s going to sit on the shelf.”
Wielbut says small changes remove steps, too: changing a drop-down to a button saves a click or two, and “when you do something hundreds of times per week, those clicks add up.” Clear messaging counts too. If the system rejects a home, saying more than “unknown system error” helps an agent fix the problem quickly instead of working through a long list of fixes or abandoning the quote.
Removing a step can mean removing a feature. Harman recalls Agency Defaults, a setting agents repeatedly asked for. “It predominantly caused confusion and friction,” he says. Agents forgot they had set it up, or didn’t tell colleagues, and were then surprised by the defaults. Rather than add more reminders, the team talked with agencies using it and removed it. “It simplified the product, reduced work for our employees, and improved agents’ ability to understand and intuit their experience when quoting with us.”
Fit also depends on the process. ACT found that 56% of agencies have no written policy on staff use of AI tools, and more than 4 in 10 rely on peer-to-peer coaching to learn new systems. Tools adopted without a defined workflow tend to work differently for every person. The ACT report puts it plainly: competitive advantage comes less from the number of tools an agency licenses than from how deliberate it is in defining and governing its workflows.
Insurance clients still want their agent. That isn’t in doubt. In the Big I study, 86% of agencies said clients prefer service through the agency by phone or other offline methods. The same study found agencies are comfortable with self-service for policy documents, billing, and claims (routine or paperwork-heavy activities) but not for buying a policy or getting a quote.
That's the role technology should play: handling the routine. Technology should return time to the conversation, the guidance, the support, and not prioritize replacing them. An agent who isn't re-keying data or chasing a fourth login has more room to explain a deductible, or to catch a coverage gap before an unsuccessful claim exposes it.
Responsibility sits with technology vendors and insurance providers, too, and not just on insurance agencies. For Openly, Wielbut says, the measure is premium growth. “Ultimately, what we build for agencies needs to translate back to premium growth.” After each release, the team tracks quote counts and conversion, and surveys agents each month for qualitative feedback.
Harman says it starts earlier, with a question teams skip: “What problem does this solve?” It sounds obvious, he says, but without it “people end up building a telephone-game translation of what was asked for.” He prefers to “stop and understand the agent’s friction point or opportunity as a business case and then collaborate on what form the actual solution takes.”
Wielbut starts with a different question for agents: “What does your dream product or system look like?” The team then turns the ideas into mock-ups, or with AI, working demos, and puts them back in agents’ hands to see whether they got it right.
The test for any team building for agents is simple: before you ask an agency to adopt something new, ask what it is replacing.
Tekippe says that’s only the first half. “Hype is not a business case. A competitor’s purchase, a vendor’s pitch, and a viral post tell you what’s popular, not what fits your agency.” The full question, he says, is what the tool replaces or improves, and how you’ll know it worked. “If you can’t answer, you probably don’t need the tool.”
Harman recommends a proof of concept or meaningful trial first because it “often turns up concrete expectations nobody thought to state when comparing tools. Tool A can replace tool B for one use case and fail to provide another.” The result is two tools or fewer capabilities.
The best technology in an agency is the kind nobody has to think about. It just clears the way for the work that only an agent can do.