Yeah it makes sense, maybe the option to have 2 different pricing schemes for each actor might make it more robot friendly while also making it reasonable for humans and developers. However I will need to do a lot of investigation into how to make this remotely profitable with PPE pricing. As for robots/AI agents they probably expect the data to be served on demand and at least for Trustpilot scraping this is simply not possible. Just starting the actor can take 20-30 seconds so it wouldn’t make much sense for example if a user is using Apify MCP to for instance get reviews from a company on demand while chatting in Claude. The mitigation of this would be to keep the actor running on standby but this also drives up the cost and price of the actor. So I think it is difficult to see it become an efficient solution for those use cases and I suspect many actors will face the same issue. There is also the 300 second synchronous API call limit so it still makes sense to async process an actor call. Most of my customers also scrape much more than just what can be done within 300 seconds.