Automation isn’t AI. Knowing the difference is worth real money.

There's a lot of "AI automation" being sold right now that is neither. Here's the line, and it's a clean one.

Automation follows rules you can write down

If a deal passes €50k, notify the VP. If a lead is from Germany, route it to the DACH rep. If a form comes in, send the email. You could describe the whole thing to a new hire on an index card. No judgment, no ambiguity, no model — just: when this, do that.

AI handles the calls a rule can't

Read this email and tell me if the buyer is annoyed. Look at this half-filled application and decide if it's worth a call. Take these twelve messy job titles and work out which are actually decision-makers. You can't write those down as rules, because the input is different every time. That's the work a model is for.

Most of what gets called "AI" is the first thing wearing the second thing's clothes.

Why the difference costs you money

A rule runs for nothing, forever, the same way every time. A model costs money on every call — you're renting it by the token. Point one at a job the other should do and you've built something slower, pricier, and, the part people miss, less reliable than the boring version.

And why it costs you trust

A rule is deterministic: same input, same output, every time, and you can prove it. A model is probabilistic — it's right most of the time, which is a different thing from right. For an email subject line, "most of the time" is fine. For deciding which leads get worked or how a record gets updated, "most of the time" is exactly how bad data gets in at scale.

So the question is never "should we use AI." It's "does this job need judgment, or just rules." Most jobs just need rules. The unglamorous truth is that a lot of "AI transformation" is really the automation you should have built five years ago, finally getting built — with a language model bolted on so it sounds like the future.

We keep the two apart on purpose. Automations are for the rules. Custom AI builds are for the judgment, and only where the judgment earns the cost. Where and why we draw that line is its own manifesto. If you're not sure which half of your stack is which, the GTM diagnostic is a fast way to find out.

Use AI where it earns its place. Use a rule everywhere else. Your bill, and your data, will thank you.