What Building Computers in 1986 Teaches About Adopting AI Today

By Win Jeanfreau

In 1986 I started Pacific Rim Computers. We built computers by hand for businesses around the Salt Lake Valley. At the time, plenty of owners weren't sure a personal computer was worth the money. Some saw it as a necessity. Others saw an expensive toy.

Today I hear the same debate about artificial intelligence. Some owners are rushing in. Others are waiting it out. Having been on the selling side of the last big technology shift, I think both groups can learn from it.

The technology isn't the hard part

In the 1980s, the machine itself was rarely the problem. What decided whether a computer paid off was whether the business had a clear job for it, like bookkeeping, inventory, or letters, and whether someone was trained to use it.

AI is the same. The tools are impressive, but a tool without a job is just a subscription. Before you buy anything, write down the specific task you want handled. "Use AI" is not a task. "Draft first replies to customer emails" is.

Start with the repetitive work

When I talk about automation in manufacturing, I put it this way: it allows the human to do the human component really well and the automation portion to do the repetitive, hot and dangerous work.

Office work has its own version of repetitive work. Sorting email, summarizing documents, formatting reports, and answering the same questions over and over are good places to start. The judgment calls, the relationships, and the hard conversations should stay with your people.

Train people before you scale the tool

Back then, a computer sitting unused in the corner was a common sight. The business owner bought the hardware but never invested in the people who were supposed to use it.

Don't repeat that mistake. Pick one or two people who are curious, give them time to learn the tool on real work, and let them teach the rest of the team. Adoption spreads through people, not through purchase orders.

Keep control of your data

Your customer list, your pricing, and your financials are the business. In 1986, that information lived in a filing cabinet or on a machine in your office. Now it can end up on someone else's server with a single paste.

Before you use any AI tool, know where your data goes and who can see it. Where you can, try local first: tools that run on your own equipment keep client information in your building.

Measure the payoff

Technology should pay its way. A computer in 1986 justified itself by saving hours on the books or catching errors in inventory. AI should meet the same standard.

Set a budget before you start. Pick a measure, like hours saved per week, response time, or errors caught, and check it after 30 and 90 days. If the tool isn't earning its place, cancel it and try something else.

Don't wait for perfect

The businesses that did best in the PC era weren't the ones that waited for the perfect machine. They were the ones that started small, learned fast, and built on what worked. The same will be true with AI.

Pick one task. Train one person. Measure one result. Then do it again.

What's the first task in your business you'd hand to AI?

About the author: Win Jeanfreau is a serial entrepreneur based in Utah.