AI is possibly as transformative as electrification in the 1880s. In contrast to AI, I doubt leaders set electricity consumption targets.

Every year, a good friend of mine who works in big tech and I spend a weekend hiking in the mountains. And during the hikes, we constantly talk about tech. Two engineers set loose, one working in big tech, the other running his own B2B SaaS company.

This year, we mainly talked about Artificial Intelligence. But despite all the hype you might see online, we talked about the leadership aspects of AI. More specifically, about the sense and nonsense of setting AI targets in your company.

The Setting

My friend in big tech enjoys a generous token budget of a few hundred dollars per day and employee.

At Yonder, the B2B SaaS company I co-founded, there isn’t a token budget; we tightly manage AI costs. I’m convinced token costs will explode in the near future. Therefore, I don’t want to make the entire company too dependent on large token budgets.

Even my friend in big tech says daily token budgets are a stupid idea. Some people don’t use them at all, while other people spend half their time building AI tools just for the sake of it.

Why Do You Need AI Targets?

Here are two questions: First, why do you need AI targets at all? And second, is a daily token budget the right metric for AI targets?

It’s quite simple. You need the AI targets to make sure even the traditionalists and backwoods people in your team start exploring AI. Here is the classic normal distribution of your team’s AI eagerness:

  • 5% of your team use AI even for tasks that could be done more efficiently without AI. These people are geeks, and they need clear leadership guidelines on what to do with AI and what not. They see the AI targets you set as a baseline, not a ceiling to aim for.
  • 90% of your team have already found some tangible AI use cases for their field of activity. Those people could happily do without AI targets. They just do with AI what is sensible, without the pressure exerted by AI targets.
  • The last 5% on your team are laggards, and they need a push to adapt to the world of AI. With all the healthy scepticism against AI, it’s impossible to claim that you cannot find an AI use case for your field of activity when you’re working in tech in 2026. This small group of people is the reason AI targets exist for your entire company.

Disillusionment. You’re bothering your entire team with AI targets when only 5% of your team need clear guidance.

And because AI euphoria is still stronger than AI scepticism, companies dole out ample token budgets and shout “adopt AI in every process and team of the organization!”

You could do much better (and much more cheaply) if you focused on your leadership role with the 5% geeks and the 5% laggards in your team. The remaining 90% could then just do their work as they already do it today – involving AI wherever it makes sense.

Conclusion: AI Needs an ROI

AI is just a tool. It’s a very powerful tool, possibly as transformative as the electrification of industry 150 years ago.

I wasn’t around in the 1880s, but I doubt leaders set electricity consumption targets, as electricity was scarce and expensive in the beginning. Rather, leaders looked at electrification as an investment in tomorrow’s productivity.

And that’s how we should look at AI targets today: AI needs to have a return-on-investment (ROI) and make us more productive in the future.

Not more, and not less.