The idea is great, but with these credits it's useless.

I have tested it and after 5 minutes the experiment is already over:
❯ Attempt #1: Using the data from a YouTube channel. Error: Too many (more than 50) documents
❯ Attempt #2: 1 file uploaded, 700kByte
❯ Then I asked about 5 questions to test the application
» Result: The credits were already used up.

❯ Now I have to wait 1 month

The product is dimensioned in such a way that even very simple applications need so many credits that it becomes extremely expensive.

If you make a mistake, you have to wait 1 month.
And it looks like it's very easy to use up all your credits: If you read in a 700kByte file, do you need 700 credits? Or if you analyse a YouTube channel that has 500 videos, do you need 500 credits even though you are only allowed to process 50 documents?

So I will be able to do another test in a month's time. Then I'll try it with a file that has 500 bytes. Then I won't need an AI, but maybe Pickaxe with Tier 2 will be able to answer more than 5 test questions.