Jelle O. If you do put a max cost to it, a few scenarios might happen.
- 1.
The prompt runs successfully without any issue (Especially if it's a simple prompt and task. Using some of the examples you've done, it does seem like putting a max of $0.001 would be fine)
- 2.
The prompt runs successfully, but it doesn't go the 'extra mile'. For every prompt you do, especially if it's web research, depending on the complexity of information, Claygent might take more steps to find or verify the information. If you set a limit, it will likely just give you whatever it can find within the cost limit. (Again, it shouldn't be a big issue if your prompt and task is simple)
As best practice, I would always recommend this:
If possible, try not to let AI do too many things. Split it into simple tasks, and get AI to just do 1-2 things per prompt. So in this case, the $0.001 limit will be fine.
If you need it to do more complex task, use something like 4o model (Has more reasoning power) instead of the 4o mini model, but it will be more expensive. And if you do use 4o model, don't put cost limits.
If you need want to control costs but also not limit what AI does, my best recommendation would be to test on 50-100 rows first, look into the average cost per row, and from there, set the limits accordingly.
There's not right or wrong answer, but those are just things to think about 🙂
