Boring 90% first: automate the 10-year-old process before the shiny experiment

The AI project that will save your business the most time is the one nobody will clap for: the quote you type out every day, the enquiry you answer for the hundredth time, the follow-up everyone forgets.

When a business starts using AI, everyone wants the exciting project instead: the showcase, the demo that impresses the room, the thing to announce on LinkedIn. I understand the appeal. But in my experience, the most valuable AI work is the repetitive process your business already runs every day.

My advice is simple: automate the 10-year-old process before you fund the new experiment.

Most Advanced Yet Acceptable

Raymond Loewy, the industrial designer behind some of the most recognisable products of the 20th century, had a principle he called MAYA: "Most Advanced Yet Acceptable." People embrace something new only when it still feels familiar enough to understand. Push too far ahead and they hesitate, however good the idea is.

I apply MAYA twice, once to where AI goes inside a business and once to what a business offers its customers. The rule is the same both times: 90% familiar, 10% new.

Inside your business: start where the volume is

The boring process has volume. It runs every day, touches most customers, and sits in the middle of your operations. A small improvement there compounds across everything you do.

The shiny experiment gets shown once and then dies, because it never touched the actual work.

That's why the demos on my Tools page are deliberately dull:

• Turning an enquiry into a quote, which most service businesses do dozens of times a week.
• Sorting the inbox by what makes money, which most owners do badly at 11pm.
• Scoring a lead before booking a call, which most salespeople do by gut feel.

Nobody gets a headline for improving these. They're where the hours are.

In your offer: too new, and nobody buys

I learned the second half of MAYA the hard way.

Mindreader, where I was CTO, built its original product to predict a prospect's communication style from a photo and from written text, with confidence rules that flag when a reading is only a guess. It is genuinely new, and that was the challenge. Buyers had no frame of reference, so they didn't know how to trust it or where it fitted in their day. Much of our time went into teaching the market what personality profiling was before we could sell it. Looking back, that product was 90% new and 10% familiar, exactly the wrong way round. It's one reason Mindreader is now moving toward AI services.

My later projects flipped the ratio:

• MummyReady used a format everyone already knows, a five-question quiz with a free gift, and put the new part underneath: automatic scoring and routing to the right advisor.
• The Modern Shrine first ran its workshops on ChatGPT, a tool attendees already trusted, before moving them to our own app once they understood the value.
• In a small first Google Ads test (23 sign-ups over 11 days), the plain "career test" wording and the more unusual framework wording converted at almost the same cost, and the familiar wording brought in more of the sign-ups. Familiar got people through the door; the difference was what they found inside.

Familiarity wins the first sale. Difference keeps the second one coming.

How to find your boring 90%

Look for the process that is:

1. Frequent: it happens daily or weekly.
2. Old: it's been done the same way for years.
3. Resented: people sigh when they have to do it.
4. Central: customers feel it when it's slow or wrong.

Anything that ticks all four is your first AI project. Then run it through Audit, Optimize, Automate, Verify before you build anything.