Feels like pricing is becoming a moving target again. Cursor’s experiments showed how fast teams will change plans the moment usage patterns shift, and LLM speed only accelerates that loop. Every new model drop forces you to rethink what’s “metered,” what’s “included,” and what users actually feel in the product.

The part that buckles first is always the billing logic. Not the API calls, but the lifecycle math behind experiments… and the experiments never stop now.

So anything that lets teams iterate without rewiring state machines every week is going to find an audience. Most people just want to ship, test, adjust, repeat, without their billing layer collapsing under the pace of AI.

Nice launch!