By Justin Sylvester, Co-Founder & CEO of PerfectQuote, who’s spent nearly 20 years in employee benefits and knows exactly how many hours brokers lose to manual data entry every renewal.
Summary: AI tools that read carrier documents faster are becoming table stakes, especially as carrier API connections return underwritten rates in minutes instead of weeks. The real advantage for benefits brokers is a single clean, unified dataset across the whole book of business, which is what lets them automate renewals, spot trends and back their advice with real analysis while client relationships stay human.
Throughout my career, I’ve spent a lot of time thinking about what actually slows brokers down. I was a broker. I made the drives, rebuilt the spreadsheets and sat across from clients who found the errors my team missed, so I know exactly what it feels like.
When I left my book of business for insurance technology, the problem I wanted to solve was a simple one: brokers were spending an enormous amount of their time on work that had nothing to do with advising clients. They spent hours entering data, rebuilding proposals and chasing carrier documents in a dozen different formats. Agencies were never short on expertise, only on hours.
That was the problem in 2016, and in a lot of agencies it’s still the problem today. Something is changing fast, though, and I think most of the industry is talking about the wrong part of it.
“We need AI.” For what, exactly?
Everyone’s building AI that could reshape civilization. You bought an AI tool that can read a quote file and draft emails. Bold strategy.
That’s the conversation I keep having. Someone tells me they need AI, I ask what they want it to do, and nine times out of ten the answer comes back to the same thing: get data into a system faster.
It’s usually not their idea to begin with. The request came from a board meeting, a competitor’s pitch deck, or a producer who saw a demo at a conference and came back convinced the agency was falling behind. What’s underneath it is almost always more specific than the technology: a Q4 that nearly broke the team, a spreadsheet error discovered in front of a client, an account manager who left and took half the context with them.
Those are real problems, and the work of reading data faster did address some of them. But that work is most of the way finished now, and the market hasn’t caught up to that yet. Agencies are still making buying decisions as though pulling numbers out of a carrier PDF were the hard part, when it’s the part everyone is about to have.
Why data extraction is becoming table stakes
For the last few years, progress in this industry has meant better OCR, smarter scraping, software that can read a carrier’s PDF and figure out that “Out-of-Pocket Maximum” and “OOPM” are the same field. That was a real problem, and a lot of smart people solved it.
Carriers are closing the gap from their side, too. They have spent years building API connectivity for enrollment and member data, and underwritten quoting is where that investment goes next. The rollout is uneven enough that it’s easy to dismiss, right up until you see one working.
Our partnership with Angle Health returns firm, underwritten level-funded rates in under two minutes, straight from the carrier’s system, with no document to read and nothing to interpret. That process used to take weeks. As more carriers build that kind of connection for underwritten quoting, nothing needs to be scraped at all, because the data arrives structured from the source.
So in a few years, “our technology reads documents faster” will land about the way “we have a website” does now.
Which raises the real question: if getting data in is table stakes, what’s actually left?
What a clean benefits dataset actually lets a broker do
Two agencies can pull in the exact same information from the exact same carriers and end up in completely different places: one with a single clean, structured view of the whole book; the other with a dozen files, three systems and one account manager’s memory.
That gap is the whole opportunity, and it has nothing to do with how the data got there.
This is not a niche concern. In EXL’s 2026 enterprise AI study, 92% of insurers said data challenges are getting in the way of their AI results, and insurance named data silos as its top barrier more often than any other industry surveyed. Only 24% considered themselves mature at managing data. Everyone wants the AI, and very few have done the unglamorous work underneath it.
A clean, unified dataset is what makes the next set of bold things possible:
- Automating the workflow around a renewal, not just the data entry inside it
- Putting AI to work across an entire book of business instead of one group at a time
- Backing up advice with real analysis, for the client with 20 employees and the client with 2,000
- Catching a trend while it’s still happening: rates creeping up in a region, a plan design losing traction, a carrier getting less competitive quarter over quarter
- Connecting what an agency knows to the rest of the tools it runs on, instead of letting it sit in a silo
None of it depends on how fast you can read a document, only on whether what’s underneath is usable.
What brokers are actually paid for
The real job of a broker is advising: understanding a group’s workforce dynamics, their claims history and their cost sensitivity, knowing when a level-funded plan is worth the conversation and when it isn’t, recommending a supplemental health strategy that people actually enroll in.
That advice is worth more every year. Mercer’s latest employer survey projects health benefit costs will rise 8.2% per employee in 2027, the steepest increase since 2003 and the fifth straight year of elevated growth. Employers said costs would climb 11% if they did nothing to manage them. Clients absorbing that kind of increase want someone who can tell them what to do about it, not a faster spreadsheet.
That kind of work takes judgment, relationships and experience. It can’t be automated, though it can absolutely be crowded out.
Getting the administrative load under control buys back the obvious thing: time. One advisory firm got days back per quote cycle and put that capacity straight into client strategy.
The bigger payoff is easier to miss. Once an agency’s data is clean and unified, the same information that used to be a dozen carrier PDFs and a handful of spreadsheets becomes something a broker can actually use. They can spot a trend, benchmark a client against similar groups, and walk into a renewal with something more useful to say than “here are your options.”
What clean data won’t replace
I want to be clear about something, because the hype cycle tends to flatten nuance.
The relationships in this industry are not going anywhere. A broker’s value has always come from the trust built with a client over years, the willingness to pick up the phone at an inconvenient hour, and the institutional knowledge no census file captures, none of which has ever lived in a spreadsheet.
Technology won’t replace that. Clients want to work with people they trust, especially on decisions that hit their employees’ health and financial security.
What technology can do is make sure the person on the other side of that relationship spent the night before thinking about the recommendation instead of fixing the spreadsheet.
What separates the agencies that win the next five years
The agencies in the best position five years from now won’t be the ones with the flashiest AI. They’ll be the ones with the cleanest data, the discipline to use it, and the sense to start getting administrative work out of their best people’s hands now.
Getting there takes three things:
- Being honest about where your team’s time is actually going
- Adopting technology even when it takes some adjustment
- Not waiting until the problem gets bad enough to become an emergency
The AI conversation in insurance is still early, and the tools are improving quickly, but the fundamentals haven’t moved. Speed wins. Accuracy wins. Trust wins. The agencies that use new technology in service of those three things are the ones I’d bet on.
The excuse that it can’t be done differently? That one is getting harder to hold onto.
________________________________________________________________________
Justin Sylvester is the co-founder and CEO of PerfectQuote. He spent years as a benefits broker before moving into insurance technology, where he has focused on eliminating the manual work that slows agencies down.