The Independent Insurance Agent Isn’t Dying. But the Job Is Definitely Changing.
There has been a funeral planned for the insurance agent for probably 25 years now, and it keeps getting postponed.
The internet was supposed to eliminate the need for agents. Online quoting was going to make shopping for insurance so easy that the middleman wouldn’t be necessary. Then, insurtech raised billions of dollars on variations of the same idea.
Now it’s AI’s turn.
A new InsuranceTalk industry report, The State of the Independent Insurance Agent 2026–2027, took a fairly deep look at whether any of this has actually happened. The answer is more interesting than either side of the argument probably wants it to be.
Independent agents aren’t going anywhere anytime soon. But quite a bit of what agents have traditionally been paid to do probably is.
We’ll start with a number that surprised me: Independent agencies placed about 62% of U.S. property and casualty premium in 2025. In commercial lines, it was nearly 88%.
Personal lines is the part I find most interesting. That’s the business where direct-to-consumer insurance should have done the most damage. It’s easier to quote, easier to standardize and consumers have been buying auto and homeowners insurance directly for years.
Yet independent-agent share actually increased from 36.7% in 2021 to 39.5% in 2025.
However, I’m not sure that is the victory for traditional distribution that it looks like at first glance.
Think about what happened during those same years. Homeowners rates sky rocketed. Underwriting tightened. Carriers pulled out of certain zip codes and states, and non-renewals became a much bigger problem.
And, brokers found that having one, two or even three carriers wasn’t always enough.
That may explain some of the increase. When insurance is easy, the consumer can probably figure a lot of it out themselves. When the market gets ugly, knowing where else to go starts having value again.
You can see a version of the same thing happening in Medicare, although for different reasons.
Medicare Advantage now covers about 55% of eligible Medicare beneficiaries, compared with roughly 19% in 2007. For years that rising tide created an enormous opportunity for carriers, FMOs and agents.
But look underneath the latest numbers.
Special Needs Plans now make up about 23% of Medicare Advantage enrollment. More importantly, around 85% of the net MA enrollment growth from 2025 to 2026 came from SNPs.
Which is a very different market than the one many Medicare agents built their businesses in. However, there is still plenty of opportunity. It’s just becoming more specialized, and national numbers don’t always tell an agent much about what is happening on the other side of the state.
A carrier can be losing members nationally while gaining them in one county. Another can look dominant nationally and be an afterthought in a particular local market. Provider networks can turn two plans that look almost identical on paper into completely different products for the person sitting across the kitchen table.
Medigap has its own version of this problem. Plan G benefits are standardized, so naturally everybody looks at the premium. But an $8 difference today may not be the most important thing about that policy.
What has the carrier’s rate history looked like? How big is the block? How old is it? What does underwriting look like if the client wants to move later?
That last one matters because a decision made by a healthy 65-year-old can look very different when that same person is 73 and has developed a couple of health problems.
This is where the technology argument gets interesting.
AI is very good at processing information, and insurance has an absurd amount of information to process. Carrier bulletins, underwriting manuals, CMS data, state filings, rate notices, provider directories, formularies, emails, spreadsheets. We have managed to create an industry overflowing with data while agents still spend an amazing amount of time trying to find basic answers.
That shouldn’t be romanticized as expertise. It’s inefficiency.
A Final Expense agent digging through ten underwriting PDFs to figure out which company will take a diabetic applicant isn’t demonstrating some uniquely human skill. Software should be doing that search.
Apparently, agents are starting to figure this out pretty quickly.
One study cited in the InsuranceTalk research surveyed 1,149 independent-agency principals and employees. Sixty-five percent had used AI for work during the previous year. The year before, it was 37%.
Agents who were using it reported saving about four hours a week.
Four hours is not a cute little productivity statistic if you own an agency.
Put ten people in the office and you’ve potentially created another workweek of capacity without hiring anyone.
And this is where I think some of the AI discussion around insurance has been looking in the wrong direction. We keep waiting for the day when a robot replaces the insurance agent.
The disruption could be a lot less dramatic than that.
One agency needs ten employees to manage a certain amount of business. Another eventually handles the same workload with eight. Or both keep ten, but the second agency writes considerably more business.
Nobody gets a headline saying, “AI Has Officially Replaced Insurance Agents.”
The economics changed anyway.
There is an uncomfortable part of the InsuranceTalk report that is worth mentioning too. It doesn’t assume agents deserve to survive just because insurance has traditionally been sold through agents.
If your value to the customer is reading a deductible off a screen, you’re in trouble.
The same goes for retrieving policy documents, explaining basic terminology or getting a quote that a consumer could get themselves in five minutes.
Even the old “insurance is complicated” defense is starting to wear thin. AI is actually pretty good at complicated.
But insurance has another characteristic that is harder to deal with: it gets messy.
The Medicare beneficiary says she doesn’t travel much. Ten minutes later you find out she spends two months every summer with her daughter in Michigan.
The life insurance applicant says he has no heart problems, but the medication list suggests there is more to the story.
A homeowners carrier technically accepts a particular risk, except anybody who has been sending business there for the last six months knows it probably isn’t getting written.
Those are not necessarily things AI will never figure out. It probably will figure out a lot of them.
For now, though, this is where experienced agents still earn their keep.
The bad part is that we are losing a lot of that experience.
The Bureau of Labor Statistics expects about 43,100 insurance-agent openings per year from 2025 through 2035. Many aren’t new jobs. They’re replacements for people leaving the profession.
Insurance has never done a particularly good job of capturing what those people know before they walk out the door.
Some of it can’t be found in an underwriting manual because it isn’t really a rule. It’s the accumulation of a thousand applications, conversations, mistakes, rate increases, carrier changes and weird cases that didn’t fit the manual.
Maybe AI eventually helps preserve some of that knowledge instead of simply replacing the people who have it. That’s one of the more interesting possibilities raised by the research.
There was one phrase in the report that was written by David Walls, an insurance broker himself, caught my attention: insurance intelligence professional.
It’s admittedly a little awkward. But the idea behind it isn’t.
For a long time, an insurance agent could create value simply through access. The agent had carrier appointments, rate books and underwriting manuals that the customer didn’t have.
That advantage is mostly gone. Everybody has information now.
The agent who matters ten years from now may be the one who can look at all that information and recognize what everybody else missed.
That person probably uses AI. They probably use better market data than today’s agent. They may handle considerably more clients with fewer employees around them. And they are going to have to know their market better than someone whose primary skill is filling out applications.
So no, the independent insurance agent does not appear to be dying.
Something else is happening instead.
The easy parts of the job are being stripped away.
What remains may actually require a better agent than the industry needed before.
