We have clear spend limits, but we started to add token budget to the "salary calculation" of our employees. We definitely differentiate between junior and senior and executive levels, what AI using and spend limit we're setting for them.

We’re a startup, so currently we don’t really have any fixed spend targets or limits as such.

If you look at how our AI budget has changed, a good example is when we were looking to hire a part-time motion designer. We were actively hiring for the role, but then we started seeing videos of Opus 5.5 creating these crazy videos, and we suddenly decided to stop hiring for that position.

What ended up happening was that our entire team of 14 people started making these 1–1.5 minute videos themselves. Within a couple of days, our two Claude subscriptions were completely maxed out and exhausted in under 2 days, even though they were supposed to last us an entire week.

So I think that gives you a pretty good sense of how quickly our AI usage—and consequently our AI spend—can change.