Three launches in ten days: how not to react

A dense fortnight of announcements produces a specific kind of bad decision. Here is what it looks like and how to avoid making it.

Since 6 November this industry has produced a downloadable reasoning model, a major point release, a new generation from Google, and a proposed rewrite of European AI law. Another release is expected next week.

Months like this reliably produce one of two bad decisions in small companies, and both are avoidable.

Bad decision one: switching on enthusiasm

Something looks better in a demonstration, and a company changes provider on that basis. Three months later they have a different set of limitations, no measurable gain, and a re-tuning bill nobody budgeted.

The defence is the method from 14 November: twenty real examples, written criteria, count the acceptable answers. If the new one does not win that test, the demonstration was selling something the test could not find.

Bad decision two: freezing

The opposite failure, and more common in Swiss SMEs. So much is moving that a company decides to wait until things settle.

Things will not settle. The 13 November post argued that the cadence has shifted to incremental improvement, which is precisely the condition under which waiting stops being rational. Waiting made sense when a generational leap might arrive; it does not when the next step is ten per cent.

What a reasonable posture looks like this week

Change nothing about your suppliers. Keep the twenty examples updated. Let any live negotiation run into January, because a more competitive market makes a January offer better than a November one.

And spend the time you would have spent reading comparisons on the internal items instead: the handover notes from Monday, or the notes in your customer system from the eleventh. Those produce value regardless of which model wins.