Off the record: what AI is doing to business models

aiblLIVE is under two weeks away. I needed to write something for this week that would be useful, but not take up too much time. So I’ve handed it over to the people we’ve been talking to.

I went back to the transcripts of calls with operators, investors, consultants, and vendors over the last few months. These are quotes we agreed not to attribute. aibl Off the Record, so to speak. Unvarnished, insightful, and occasionally disturbing. All I’ve done is clean up the language and remove any identifying information.

Once I started digging, I found quite a few I think you’ll find interesting and helpful. So I’ll split them over the next two weeks. In this edition we look at business models, ROI, and the future. Next time we’ll share quotes about operations, tech, and teams.

On business models, sales, and products

Our whole model is selling time, and the time is disappearing. Within five years I expect we’ll price almost everything on outcomes or fixed fees, and the partners who can’t make that shift will struggle. And when I say five, I think I mean three.

We’ve started packaging things our consultants used to do by hand as subscription tools. It’s lower margin but we can sell it to companies that could never afford us before.

If a customer can ask an AI assistant to compare suppliers and place the order, what am I for? I think the answer is the relationship and the problems when things go wrong, but I’m not certain that’s enough.

One of our editors used AI to code a product that pulls together all of the public and semi-public data for the sector, and gives our subs the ability to play with it, query trends and create custom reports. Again, he’s an editor and he coded this himself in a month. It’s the first thing our marketing leads with and sales people show.

My advice to anyone is don’t announce AI to customers. Talk about new features, speed and quality. Keep the focus on you.

Prospects now turn up having asked ChatGPT to compare us with three competitors, and some of what it told them is wrong. Half the first meeting is correcting the record, but in general they’re better prepared. It’s harder to manage the sales conversation overall.

Our customers’ procurement teams now benchmark our prices or at least think they do. Any price rise beyond inflation is on hold unless we show new value, and I think that’s probably fair.

To be honest I think most customer service complaints and suggestions were read, but didn’t turn into more than one-off changes. Now we summarise and plot them against capacity. Three of our last four product advances came from customer complaints we’d never properly heard.

On funding, ROI, and reporting

The first time I put AI in the board pack, I reported our active Copilot users. That didn’t last long. We’re not at a hard financial number, but I’m able to report tangible things that changed in the business. Last month I talked about a role my marketing lead didn’t need to fill, because AI had picked up the requirement.

Our operating partner circulates a quarterly table of portfolio companies ranked by AI adoption. It’s motivating but slightly mad, because half of what’s being counted is people logging in.

Bid writing was the first place AI paid for itself, no contest. We went from three people working weekends on a framework submission to one person working normal hours, and our win rate held.

We give each department £15k a year to try things, no business case needed, with one rule: report back what happened, including if it failed. About a third of those experiments turn into something real.

AI has turned what used to be a one-off software purchase into a monthly bill that grows with usage. My board understood the old model and they’re nervous about a cost line with no ceiling. I’m trying to guide them to an understanding that now is not the time to be penny-wise and pound-foolish.

We built our own quoting tool and only realised last month that some of it qualified for R&D tax relief. Nobody in the AI conversation was talking to our tax advisers.

Our first project stalled, and our second took 10 months to show proper returns, and I nearly lost my nerve. The most recent one took a quarter and it was revenue positive almost immediately, because we’d learned what to do and what not to do.

We’re getting real returns, but they show up as things not happening: errors not made, overtime not paid, staff not leaving. It’s harder to put on a slide than revenue, but it’s in the numbers if you look.

Our biggest mistake was not measuring how long things took before we introduced AI. Now everyone feels it’s faster but we can’t prove it.

On the future

I advise an investment fund when they are looking at businesses in the insurance industry. A month ago I did a deep dive on a start-up that services manufacturing companies. It’s specialised work and I went in assuming they’d have good tech, but not have the systems in place to overcome the moats of the established players. Bottom line: they have a team of five and they’ve created the best solution in the industry in six months. They have one person who knows the business and four coders. It’s only possible because they were AI from the jump. Amazing.

My daughter asked ChatGPT for the best suppliers in our niche and we weren’t mentioned, but two competitors half our size were. That got me more worried than any analyst report. No offence.

I came into a plan to double revenue by 2030 with roughly the same number of people. I’d have called that fantasy when I was hired, and now I’d say I’m 60% sure it’s not.

I think we (financial services) end up looking like an hourglass: senior people with judgement at the top, a lot of operational staff at the bottom, and much less in the middle. That middle layer is where most of my managers sit, which is a conversation I haven’t had yet.

For our 30-year history, the big players beat us on scale. Now our 300-person firm can do what used to take 1,500, the thing that matters is how fast you can change, and we’re much quicker than them.

I’m planning to hand over in four or five years and I’m not sure the business I’m handing over will look anything like the one I built. My successor needs to be comfortable running something I don’t fully understand.

What we took from it

What strikes us reading these back is how little of it is about technology. The quotes that stayed with us are about pricing models, board conversations, succession, and the middle of the org chart. Nobody is wrestling with which tool to buy. They are working out what their business is for once the thing it sold is cheap. That is the conversation we built aiblLIVE around, and it is the one happening in the hallways rather than on the stage.