What a 1950s teashop can teach us about AI

More than 70 years ago, a British teashop chain showed that new technology creates most value when it changes how the whole business works. Anthony Impey argues that we need the same approach to AI today.

It is hard to believe that the world’s first computer built specifically to run a business was created not by a technology company, but by one best known for tea and cakes.

In 1951, J. Lyons & Co ran hundreds of teashops, alongside bakeries, hotels and a food-manufacturing operation. Coordinating all of that required the company to process information about orders, stock, production, deliveries, costs and employees.

Its answer was the Lyons Electronic Office, or LEO, the first computer designed specifically for business operations.

LEO was initially used to calculate bakery costs. It went on to process payroll, manage orders, plan production and coordinate deliveries. Lyons even used spare computing capacity to undertake work for other organisations, including Ford – in some ways, an early version of the service AWS provides today.

But the most important part of the story is not that a teashop business built a computer. It is how Lyons decided what to do with it.

That provides a valuable lesson for businesses considering what AI could mean for them today.

Lyons did not begin with a new technology and then search for somewhere to use it. It began with a detailed understanding of how its business worked.

Its managers knew how information moved from the teashops to the bakeries, how orders shaped production, how products were allocated to delivery routes and how this affected costs. LEO was introduced to improve those systems and processes, rather than make one isolated task a little faster.

And yet, nearly 75 years later, too many businesses are approaching AI from almost the opposite direction. They start with the technology – buying licences, running pilots and encouraging people to experiment – rather than asking where the business is wasting time, losing money or missing opportunities to grow.

That can create a lot of activity without changing very much. An AI tool might cut ten minutes from producing a quote. But if the quote still waits two days for approval, has to be entered manually into another system and reaches the customer after a competitor’s, the business has gained very little.

The greater prize lies in improving the whole chain of activity: responding to an enquiry, producing a quote, planning the work, delivering it and then getting paid.

So, the starting question should not be: where can we use AI? It should be: where could better information, faster decisions or fewer handovers improve the performance of the business?

Answering that requires the same discipline Lyons brought to LEO. Leaders need to understand how work moves through their organisation, involve the people who know those processes best and be clear about what needs to change.

The evidence should then show up in the performance of the business: happier customers, healthier margins and more capacity to grow.

The real lesson is that Lyons did not use LEO simply to make the same business run a little faster. It used it to change how the business worked. That remains the test for AI today: not how much of it you use, but whether it helps you run a better business.

The post What a 1950s teashop can teach us about AI appeared first on Elite Business Magazine.