Hey guys, I’ve been thinking about how usage-based pricing is becoming more common, especially for companies who want to monetize AI products. It makes sense, right? Pricing customers based on how much they actually use feels fairer. But I’m wondering, how do you balance that with the unpredictability of AI resource consumption? Are there any frameworks or models that have worked for you?

We've discussed this a while back, but history saving costs money on slack so here's what I can remember from the chat: • If doing PayGrow, have spending caps • Actively choose whether exceeding the cap shops service or just slows it down • PayGrow works well for new tyre kickers and small or periodic users, but you will likely also want tshirt costs for companies who want a predictable monthly cost • You'll also likely want both options if you move into enterprise support plans, as revrec for investment is often based on committed spend • I personally think OpenRouter's approach of buying credits easily and warning when they're running low is good.