AI for client relationships: remember every detail.

I had lunch recently with a friend who sells insurance. Good at it, too - the kind of person clients actually want to spend an hour with. I asked how he was using AI. "Barely," he said. He'd once built a pricing table in Gemini, which laid the options out cleanly for a client, and that was about it. Then the line I hear constantly about AI for client relationships: "I don't find it that useful. I still need to go and see the client."

He's right. And it's worth saying plainly, because most of the pitch around AI gets this backwards. Seeing the client is the job. That is the part that closes business, and no model is going to do it for you. If anyone is selling you AI as a substitute for the lunch, walk away from them.

The meeting was never the part AI replaces

I asked what a typical client lunch actually looks like. Ninety per cent of it, he said, is personal - kids, holidays, tennis, the usual. Maybe ten per cent is insurance. He said it slightly apologetically, as if that were a confession.

It isn't. In a relationship business that ratio is not a bug, it is the entire point. People buy from someone who knows them, and the chitchat is the product. If AI tried to sit in that chair it would ruin the exact thing that makes the meeting work. It is the same boundary I set out at length in the honest answer on whether AI replaces your job: the judgement and the relationship stay yours, and the typing does not.

So no - AI does not go to the lunch. It does everything on either side of it.

What AI for client relationships actually does

Here is the part he was missing. That ninety per cent of "just chitchat" is the most valuable thing said all afternoon, and most people let it evaporate on the drive home.

Picture a client - call him Brian. Over lunch you learn he has two kids, the elder just started at a school in Kowloon Tong. He is on a particular medical plan and is not looking to change it this year because he only just renewed. He plays tennis on Saturdays. None of that is a transaction. All of it is the relationship.

After the meeting - not during it - you spend ninety seconds turning your memory into a short, structured note. AI does the tidying: it takes your messy voice memo or three scrawled bullet points and files a clean record. Brian: two kids, elder at the Kowloon Tong school, on his current plan, not switching this cycle because he just renewed, keen tennis player. Six months later, before you see him again, that note is waiting. You ask how his son is settling in and whether he is still playing on Saturdays.

To Brian, you are the adviser who remembers everything. That is not a trick. You did have the conversation - AI just made sure you didn't lose it. The prompt that does the tidying is about four lines long, and there are more like it in the free prompt library.

AI won't remember your clients for you. It makes sure you never forget them.

Now multiply Brian by a hundred and fifty. Nobody holds that many relationships in their head at that level of detail. The people who appear to are running a system. It used to be a shoebox of index cards; now it is two minutes of dictation and a model that keeps it straight.

Before the meeting, and finding the next one

The same logic runs forward. Before a catch-up, AI can pull your last few notes on someone into a thirty-second brief, so you walk in already caught up instead of half-remembering on the way over. Further up the funnel it helps you build a pipeline and work out who to see next - the sourcing and the prep that most people do badly, because there is never enough time. That is the quiet win: not doing the meeting for you, but handing back the hours around it so you can do more meetings, and do each one properly. It is the same arithmetic that makes automating the Excel and PowerPoint busywork worth doing - the hours come back from the admin around the work, not from the work itself.

The honest limit

None of this builds the relationship. Warmth, judgement, the read on whether today is the day to raise business or absolutely not - that stays entirely human, and it is most of the value. Feed AI nothing real and it will happily invent a tidy note about a conversation that never happened, so the person still has to have actually paid attention. The tool is only ever as good as the human who showed up.

There is a second caveat worth naming, since we are talking about other people's personal details. What goes into that note about Brian is his data, not yours, and where you type it matters - a consumer chat window and a properly set-up work account are not the same thing on that question.

But that is the reframe worth sitting with. AI in a relationship business is not there to make you less human with clients. It is there to give you the bandwidth to be more human, more often, with more of them - and to never again be the person who forgot that Brian plays tennis.

Where this starts

If you carry your client relationships across your head, your phone and a scattering of half-finished notes, that is the normal starting point, not something to fix before we talk. A working session starts exactly there - your real clients, your real book - and rebuilds the habit around them. The lunch stays yours. Everything around it gets easier.