Bottom line up front: There is a difference between AI-first and AI-forward. AI-first is a procurement decision: rip out the old systems, replace your people with software, become a “tech-enabled broker.” It fails in the field. AI-forward is an operating model decision: keep your people, fix your workflow with the SSAMM Method™ (Simplify, Systemise, Automate, Monitor, Manage), and use AI to remove the grunt work strangling your account executives, renewals team, and compliance function. Then redirect the recovered time into the two things AI cannot do: the broker relationship with the client, and the broker relationship with the underwriter.
Most UK insurance brokers I meet have bought ChatGPT licences for the office, watched a webinar, and then quietly gone back to doing the job the way they did it in 2019. The difference is that now they feel guilty about it. That guilt is not the problem. The problem is that buying AI licences is not the same as running an AI-forward broker, and most of the industry is mistaking one for the other.
I am writing a book about this called The AI Forward Broker. I am also speaking about it at a UK broker conference in June. So this article is the public preview of an argument I am going to be making for the rest of the year, and it is the first in a twelve-week series for brokers, MDs, and account executives who are tired of being told AI will save them and would quite like someone to show them how.
Let me get the bottom line up front.
There is a difference between AI-first and AI-forward. AI-first is what the consultancy slides sell you. Rip everything up. Replace your people with software. Become a “tech-enabled broker.” It is rubbish, and worse, it is rubbish that fails in the field. AI-forward is what actually works. Keep your people. Fix your operating model. Use AI to remove the grunt work that is currently strangling your account executives, your renewals team, and your compliance function. Then redirect that recovered time into the two things AI cannot do, the broker relationship with the client, and the broker relationship with the underwriter.
That is the entire thesis. Everything that follows is plumbing.
Why “AI-first” keeps failing in broker offices
I have sat across the table from at least a dozen broker MDs over the last two years, and the pattern is identical every time.
Year one, somebody buys ChatGPT Team. Account executives are told to “experiment.” A few of them produce funny limericks about the office cat. One of them generates a half-decent draft email to a client. The MD declares progress. Year two, a “Head of Innovation” is appointed, usually the MD's nephew or a 27-year-old who once worked in fintech. Year three, a six-figure consultancy invoice arrives proposing an “AI transformation roadmap.” The roadmap recommends platform decisions that need to be made by people who are too senior to know what an account executive actually does at 9.30 on a Monday morning.
Year four, nothing has changed except the broker has spent £150,000 and produced a PowerPoint deck.
I am not making this up. I have got the deck on my desk. Two of them, in fact.
The reason this fails is simple. AI-first treats AI as a procurement decision. AI-forward treats it as an operating model decision. One is a software contract. The other is a way of running the business. Most brokers do not have an AI problem. They have an operating model problem, and AI is exposing it.
The AI-forward broker, defined
An AI-forward broker is one where every part of the workflow has been audited for two questions.
- Does this task need a human relationship, or does it need a human signature?
- If it only needs a signature, why is a human still doing it?
That is it. That is the whole audit. You walk through your renewal process, your new business process, your claims handling, your MTAs, your underwriter submissions, your compliance file checks, and you ask those two questions at every step. The answers will surprise you. Most brokers find that 60 to 70 per cent of the daily activity in their office is signature work, not relationship work. AI does signature work for pence. Humans do relationship work for hundreds of pounds an hour. The maths is not subtle.
Once you have done the audit, you apply what I call the SSAMM Method™. Simplify, Systemise, Automate, Monitor, Manage. The order matters and most brokers get it wrong by skipping straight to Automate. You cannot automate a process you have not simplified. You cannot systemise a process you have not standardised across the office. And you absolutely cannot monitor a process where the inputs are coming from six different versions of an Excel template that Dave knocked together in 2017.
Simplify first. Then systemise. Then automate. Then monitor the numbers. Then manage by exception. The brokers who run their offices that way are the ones quietly outperforming everyone else, and most of them will not tell you, because their advantage is real and they do not want to give it away.
The three places I see AI actually working in UK broker offices
I will write a full piece on each of these in the coming weeks. For now, the headlines.
Renewals. The renewal book is the most underloved asset on a broker's balance sheet. Most brokers treat it as admin. Pull the data, send the letter, chase the response, log the result. AI flips that into a triage exercise. Which renewals are at risk. Which are over-priced for the market. Which can be auto-bound. Which need a phone call from a senior account exec. The AI does not make the decision. It tells you which ones need your attention. That is the whole game.
Underwriter submissions. The brokers who win in a hard market are the ones whose submissions are clean, complete, and easy for an underwriter to say yes to. AI cleans submissions cheaper and faster than a junior account handler can. It catches missing information, formats the data the way the underwriter likes it, and writes the cover note in the house style. Submission quality is the only number that matters in placement, and AI just made it cheap.
Compliance. Consumer Duty, the FCA's vulnerable customer guidance, suitability assessments. All of it sits on top of evidence trails most brokers cannot produce on demand. AI builds the evidence trail in the background while the account exec is doing the actual job. By the time the audit comes round, the file is already there. Sleep at night, do not cut corners.
The honest aside
I got this wrong for years. When I first started talking to brokers about AI back in 2023, I was in the AI-first camp. I thought the answer was to rip out the old systems and replace them with something modern. The brokers who listened to me wasted money. The brokers who ignored me and just used AI to fix the bits they hated quietly grew. The market is the teacher. I have updated.
There is a version of this story where the AI-first consultancies turn out to have been right, and the brokers who hung on to their operating models look like the buggy whip manufacturers of 1908. I do not think we are in that version. The reason is that broking is, fundamentally, a relationship business. The client buys the broker, not the policy. The underwriter trusts the broker, not the broker's tech stack. AI does not change either of those facts. It just changes how much time you have to invest in them.
What you do this week
If you are a broker MD reading this and you want one thing to action before Friday, here it is. Walk into your office tomorrow morning. Sit at the desk of one account executive for thirty minutes. Watch what they actually do. Do not talk. Just count.
Count how many times they switch screens. Count how many times they retype something that was already typed somewhere else. Count how many emails they send asking for information that was attached to a previous email three days ago. Count how many minutes they spend in the broker management system versus on the phone to a client.
You will be horrified. That is the start of the audit. Once you have seen it, you cannot unsee it, and you will never again accept “we already use AI” as an answer when what you actually meant was “we have a ChatGPT subscription.”
Simple, specific, scored. That is how you become an AI-forward broker. Everything else is noise.
Next week, we go inside the renewal book. I will show you what £180,000 hidden in the renewals process actually looks like, and the four-step fix that gets it back.
Pick one part of your operating model. Look at it honestly. Tell me what you find.
This is article one in a twelve-week series for UK insurance brokers, drawn from my forthcoming book The AI Forward Broker. If you want to get the rest as they publish, connect with me on LinkedIn, or book a fifteen-minute discovery call and let's talk about your operating model.
