Intelligence-First

The most common mistake I see in companies adopting AI is treating the purchase as the strategy. They choose a tool, put a team in charge of it, and expect the advantage to follow. I understand the appeal. Buying something is a decision you can finish.

Changing how a company works takes longer, and it’s harder to put in a budget. But that’s where I think much of the value will come from. I’ve started calling that approach intelligence-first: building the ability to use these tools into the way the business operates.

There is a precedent in the move to digital. Putting an existing business on a website was useful; rebuilding around what the internet made possible could change the business itself. Ride-hailing is an obvious example. The app connected the customer, driver, dispatch, and payment in a way that changed what people expected from the service.

I think AI creates a similar opportunity, though I don’t know how quickly it will spread or which companies will make good use of it. The bubble worries me. A collapse could interrupt funding and kill vendors that customers depend on. It would still leave businesses with useful capabilities and decisions to make about them.

What changes after the purchase

Consider a law firm buying a research assistant and a drafting tool. Both may be worth the money. Its competitors can buy them too, so owning the software alone won’t distinguish the firm for long. What matters is how it uses the tools: where they fit into a matter, what information they can reach, and who reviews the work.

That is the level at which I want companies to think about AI. A tool needs a place in the process, with someone responsible for its output. If it sits off to the side while people continue doing everything the old way, the purchase hasn’t accomplished much.

Software is part of this. For much of my career, custom development was expensive enough that many companies had little choice but to adapt to what they could buy. Small, purpose-built tools are becoming more practical. Developers like me are among the people whose work these tools have made cheaper. That opens up projects that previously couldn’t justify the cost, although someone still has to gather requirements, maintain the software, and decide whether it should exist at all.

The opportunities extend beyond the technology department. Take a finance team buried in closing the books and producing reports. If some of that work takes less time, what could they investigate? There may be a question they haven’t had time to answer, or a view of the business they could build for another department. Giving people room to use their judgment is one of the reasons I’m interested in this. It requires a decision about what to do with the time saved; it doesn’t happen automatically when a tool arrives.

The asset underneath all of this is a company’s particular way of working. People often describe it vaguely: “our service is better.” I’d like them to be able to explain why. Which decisions do their people make well? What do they know about the customer that changes the service? Where does the process reliably produce a better result? Those are things a competitor can’t reproduce simply by buying the same subscription.

Start with the process you have

A difficulty I encounter is that companies don’t always know how their own work gets done. There is a clean version in a presentation, and then there is someone copying data into a spreadsheet because accounting never got a seat in the system.

That workaround matters. It tells you something about access, cost, and the gap between the intended process and the one people can use. Describing the actual work gives you somewhere to start. Without that, it’s hard to judge either the cost of a problem or the value of fixing it.

There’s also resistance that has little to do with the tool. I’ve watched executives push back on technology when they realized the visibility it provided would expose their own decisions. Better reporting can make it harder to dismiss a missed target or explain away a forecast that someone overrode. A regular review of the numbers asks something of the people in charge, too.

That doesn’t make every objection self-interested. Costs, reliability, and disruption are reasonable concerns. But buying software won’t resolve a disagreement about who is accountable, and it’s expensive to discover that after the implementation.

What I’m betting on

My bet is that companies will get more lasting value from improving their own operations than from chasing each new tool. The useful work is specific: understand a process, decide what needs to improve, and build or buy what helps. Then check whether it helped.

That may start with a modest internal tool or a report someone has wanted for years. I’d rather see a company get one of those into regular use, learn from it, and keep going than declare itself intelligence-first on the strength of a purchase.