Hey folks, I am excited about the potential for AI integration however I am stumbling a bit with getting things to work. I have a business account with ChatGPT. In the app I can run unlimited queries (i think). But I am having issues with slow response and just basically hangs on the prompt. Sometimes i get that i have exceeded my quota, and sometimes it works. As mentioned, I don’t know that I have a limit, but that does not mean I don’t. But the interaction with it seems sporadic and nonsensical. One example is that the Chat box on the right side will do one thing but when i open the chat pop up i can get different response.
Is anyone having consistent luck here with ChatGPT or with another AI like Claude?
One example is that the Chat box on the right side will do one thing but when i open the chat pop up i can get different response.
The Chat assistant is accessed via the Chat popover or the Chat inspector. They are independent of each other, by design.
I have a business account with ChatGPT.
But do you have an API key ?
Is anyone having consistent luck here with ChatGPT or with another AI like Claude?
Of course, and not only with those two AI providers but in some cases with local models as well.
But the interaction with it seems sporadic and nonsensical.
Please provide more information on what you’re doing and prompting. And remember, LLMs aren’t a static repository of information but instead generates responses on demand. You should expect a certain amount af variability in the replies, even with the same prompt and same provider.
Did you read the Getting Started > AI Explained section of the built-in Help and manual? This section explains basic of AI and the way it has been integrated into DEVONthink. The subsequent AI in Practice section provides some more practical information on its use and learning to chat with AI in DEVONthink.
Forum post viewed and AI-prompted directly in DEVONthink with Claude Sonnet 4. Lower-tiered models, e.g., Haiku 3.5, may not be able to accomplish this, but testing of models from other AI providers would need to be done to determine this.
i was not certain of their independence but got that impression, however, sometimes i can make the popup work with the same prompt that fails in the inspector.
I do have an API key. I did read at least portions of the manual. Currently i have run simple prompts like the batch processing rename of the file as well as requests for summaries etc. This is all with GPT as mentioned. I was wondering if I should pay for another account to test if this is an openai issue or what. .
The discrepancy could be network issues, server responses, or access to a specific model. The latter is quite possible. I prefer Claude Sonnet 4 for anything requiring a bit of effort but it can be overloaded at times. This is notable when using the Claude.app.
What GPT model have you set as your default in the AI > Chat settings?
like the batch processing rename of the file
A built-in batch configuration, using what model?
as well as requests for summaries etc.
Using the Chat assistant or Edit > Summarize via Chat or Tools > Summarize Documents via Chat?
I am wondering if there is an issue with having a personal account that is free and a paid account. I have the paid team ChatGPT but it seems that my api key is perhaps out of my personal account. If that is the case, maybe the paid account does not provide an API key? I am thoroughly confused because when i go to my account i have personal and the paid. but when i go to create the API it seems like its one and the same.
I have tried a variety of models including the mini and nano.
i think it is an account issue maybe or maybe there is no such thing as a paid account api key.
this seems to indicate that i have an API from my unpaid personal account. but it also shows no activity. so not sure why its sometimes telling me i am over the quota.
