Currently, the assistant in the Create panel primarily provides the description, the logo image etc. Nice, but these are easily done by the developer anyway. However, it does practically nothing for the Actions and if you go to the Configure panel, to add an Action manually, the panel is quite buggy. If you attempt to edit an existing Action the edits are not saved. You have to copy the Action content, delete the Action, and re-paste it back to a newly created Action and do the edits there. The most messy part of defining a custom GPT is getting the communication through the API endpoint to work properly. Currently, the documentation and examples are minimal and not very helpful, and a lot of us are struggling to get them configured properly. This combined with the tight usage caps imposed, makes the process of debugging the communication of custom GPT to an outside API endpoint extremely painful because it interrupted all the time. Also the usage wasted on debugging eats away usage from the account for other uses of the GPT4. So my suggestions: the most useful things that the assistant in the Create panel could do is to ask the developer to define the URL, the communication and authentication features in simple language and then based on the answers to these questions create the actual openai schema that goes in the Action’s config panel. That would really be saving a lot of us major pain and would free GPT4 bandwidth wasted on this tedious debugging process of the Actions.