Pipeline isn't an activity. It's an asset.
Read →
By appointment
May, 2026
Article No. 02
AI can write your posts. It can't make you credible.
Why credibility is still a human product. The one thing the model on the other side of the prompt cannot supply.
Most brokerages don't lose business because they lack expertise. They lose because that expertise never shows up where the people thinking about hiring them happen to be looking, and the gap between knowing something and being seen to know it is, increasingly, the gap that matters.
The problem, in other words, is visibility, and on the surface artificial intelligence looks like the answer to it. Content gets faster, volume goes up, costs come down, and the production line of marketing seems to have caught up to the rest of the business.
It has not.
In insurance, where what a firm publishes is read at the same moment the firm itself is being assessed, visibility without credibility does not fail quietly; it works against the firm producing it.
What a prospect evaluates when they pull up your website or your team's profiles is not only how much you publish, or how often. It is something quieter and harder to fake, and it is the one thing the model on the other side of the prompt cannot supply.
Social media doesn't create demand. It converts it.
Before any conversation about AI, it is worth being clear on what social media actually does for a brokerage, because almost every article written on the subject gets this part wrong. Social media in this market does not generate cold demand. What it does, when it works, is convert warm intent.
A prospect hears the firm's name from a referral or a passing comment at an industry event, and within a few minutes they are on Google, then on the firm's website, then on LinkedIn, then on the individual profiles of the people who would be sitting across the table. The whole assessment takes about seven minutes, happens silently, and is mostly finished before a call is ever made.
This is not theory. It is how serious buyers behave, and it is also how the producers worth recruiting evaluate brokerages long before they entertain an offer. What looks from the inside like marketing is, from the outside, something closer to evidence: that the firm understands the work, understands its clients, and is worth handing over the part of a business that, when it goes wrong, ends careers.
The job of a social presence is not to argue for the brokerage. It is to remove every quiet reason a thoughtful buyer might decide not to call you, which is exactly the work AI is least equipped to do.
Compliance is not a prompt problem.
In insurance, every piece of content sits inside a regulatory frame, and the question of what may be said, by which licensee, in which jurisdiction, with which qualifications attached, is not an afterthought to the work but the architecture of it.
What you say matters. How you say it matters more. What you choose not to say sometimes matters most of all. AI understands none of this. It has no view on where education becomes advice or where a confidently worded sentence crosses a regulatory line, because it has no stake on either side of that line. What it does instead is hedge into safe, generic language, the kind that sounds compliant and credible everywhere because it is compliant and credible nowhere in particular. To a sophisticated buyer, that reads as the absence of authority, which is the wrong signal for a firm whose product is credibility.
Real compliance is not the avoidance of mistakes. It is the exercise of judgment inside a set of constraints, and judgment requires accountability. AI cannot lose its license. It cannot face a regulator. It cannot answer for what it published, because the downside, by design, lives somewhere else, which is another way of saying it lives with you. Every serious operator in this space, ourselves included, runs content through human review before anything goes live.
Your social presence is representation, not content.
Most people misunderstand what social media is. They think it is about posting, when in fact it is about representation; the public projection of the firm into the moments before contact. Every post a brokerage publishes answers, whether the firm intends it to or not, the only question a prospect is actually asking: what would it be like to work with these people? The answer is not in the information. It is in the tone, in how the firm explains complexity, frames risk, handles disagreement, and speaks to the clients who already trust it.
AI flattens all of that. It produces language that is technically correct and culturally empty, and what gets lost in the flattening is the very thing the firm was hoping to communicate. If the content does not reflect the brokerage, how the producers speak, and how the team actually operates, it does not build trust. It erodes it, slowly, in the part of the prospect's mind they cannot quite articulate, until at some point the call quietly fails to get made.
Simulation is not the same as expertise.
This is where the deception is most useful and most dangerous. AI sounds knowledgeable. It uses the right vocabulary, structures ideas cleanly, and explains technical concepts with the calm authority of a textbook, but it does not actually know anything. It aggregates patterns; it does not synthesize experience. In insurance, that distinction is the difference between a colleague and a stranger reading from a script.
Real expertise shows up in the claim that didn't go the way anyone expected, in the coverage gap that nearly got missed, in the renewal that almost slipped over a single line of policy language nobody flagged until the third reading. It shows up in the texture of how a producer describes a market, the quiet authority that comes from having been in the room when things went sideways. None of that is in the model. The advantage in this market has always come from knowing the rooms you sell in, the clients you serve, and the language that lands inside them, and AI has never been in those rooms. It borrows everyone else's language, badly.
Credibility lives in context.
Good brokerage content is not static. It responds to markets, to regulation, to current events, to the small shifts in what underwriters are tightening or what clients are quietly worrying about this quarter. AI is always slightly behind that present moment, and even when its training is current, it lacks the conviction to take a position; it hedges and generalizes in a way that reads, to anyone paying attention, as the absence of a point of view. Credibility comes from being willing to be specific, and specificity comes from being inside the conversation rather than reassembling its surface from fragments.
Judgment is the product.
What to publish, what to leave alone, when to weigh in, when to stay quiet: that is the actual work, and it is the work AI cannot do. The model has no stake. There is no downside if it gets something wrong, no reputation in the balance, no client relationship at risk, no Monday morning phone call from an account that saw something it did not like. Without stakes there can be no real tradeoffs, and without tradeoffs the output reverts to whatever the training data considered safe, which is, by definition, what everyone else has already said.
A senior marketer inside a brokerage knows when an opinion is worth publishing and when it is not. They know which carrier relationships to leave alone, which hard market commentary will read as insight and which will read as complaint, and which line a regulator will let pass and which line they will not. They are not following a rulebook; they are reading a room, and the room is built from years of being in it. In a professional services firm, that judgment is the product. Everything else is packaging.
Where AI actually fits.
None of this is an argument that AI is useless. Used in the right place, by people who already know what they are doing, it is a real tool. It can draft a first pass quickly, structure an argument, stress-test a headline, compress an hour of work into twenty minutes when the work did not need to take an hour to begin with. Used that way, it makes good marketers faster. Used as a replacement for the marketer, it makes the firm's voice quieter, blander, and harder to remember. AI should support the thinking. It should not replace the thinker.
The brokerages that win will get this right.
The argument here is not against AI. It is against losing what makes a firm distinct in the rush to use it. The advantage in insurance has never been information; everyone has access to that. The advantage is interpretation, the judgment and experience and perspective that come from the view inside fifteen years of policy reviews, renewal conversations, and claims that didn't follow the script. That is what clients are buying when they choose one brokerage over another, and it is what prospects are looking for when they pull up the firm online before deciding whether the call is worth making. AI can help a brokerage show up. It cannot, on its own, make the firm worth choosing. Not yet.
By Appointment
Become the brokerage
they recognize.
Fifteen minutes on the phone. No slides, no pitch, no pressure. Just a conversation about what it would look like to be seen in your market.
Insurance Social
In Pursuit of Recognition