Getting started
Don’t start with an AI strategy. Start with the inbox.
4 min read · 10 July 2026
“We really should do something with AI too.” The sentence comes up in almost every company these days, and usually a big document follows it, a strategy, a tool evaluation, a project proposal. At our place, something smaller follows it: a program that runs every morning at 8:45.
What happens at 8:45
The program reads our mailbox, in both directions. For every open contact on our sales list it checks whether a reply has come in and what we last wrote ourselves, records both on the contact’s card, and then rewrites the day’s list: which replies are due today, where to follow up, where there’s been silence for two weeks. The list that used to form in your head every morning while scrolling through the inbox is sitting there finished before the coffee is done.
Two quirks of it are deliberate. The program runs on our own machine, with no cloud service in between. And it never invents data: if it finds no documented email address, the field stays empty instead of a plausible one being guessed in.
Why this job and not a bigger one
Because it has three properties that carry a first use of AI. It comes up every day, so every hour of work on it pays off daily. Its result can be checked in a minute, one glance at the inbox is enough. And when it gets something wrong, nothing bad happens; one line on a list is wrong, that’s all. Sorting work, not deciding work.
The usual first attempt is the opposite: the customer chatbot, the automatic quote, the big visible thing. There, the result is hard to check, the damage in case of error is real, and because nobody can watch the machine work, nobody trusts it either. Many AI projects don’t die of the technology, they die of this missing trust.
The strategy doesn’t come before the first running thing. It grows out of it.
What the small thing turns into
After a few weeks with a helper like this, you know two things from your own observation: what the machine can do reliably in your business, and where it gets things wrong. With these two observations, the strategy conversation becomes a different one. It talks about your operation instead of brochures, and you pick the next automation where the first one is already carrying weight.
How we work AI consulting
In a consultation we first look for your 8:45 moment: the sorting work that comes up every day in your business and whose result is quick to check. That’s what we build first. We talk strategy as soon as something is running that it can be pinned to.
Write to us and tell us which task eats time every morning at your place. That’s usually the start.
“We really should do something with AI too” is a good sentence, by the way, it shows the curiosity is there. It just needs a different continuation than a document. At our place, it hangs on a time of day, and the workday has since stopped starting with scrolling and started with a list that’s already there.