Time out issues

Hey there,
I recently bought a licence and am trying out Chat Assistant Pro. So far, i am happy with it, but i encountered some issues.

I followed your video about onboarding, connected to open ai api ( which is payed ), i created assistant, added knowledge from the page and added custom tools.

I also set to use the file search option with one assistant, leading him to use a pdf file as a source for information. ( Since i couldn't find any direct instructions how to do that, here is what i did:
1. I uploaded one file via the files submenu
2. In one specific assistant, under the File Search Tab, i clicked on yes, clicked on create vector store and selected the file i had previously uploaded.
3. Finally I set custom tools to also search knowledge base.
When I would click on save and close, the first time i got an error message ( which unfortunately i did not saved or screenshot )
after re-doing the same, it saved without error message. 
I clicked on vector store and the vector store for that specific file was created twice, both with status complete.

The first tests then on the frontend were great, but after three following questions, the typing animation would not stop and after a while, i'd get timeout errors like this:
Idle timeout reached for "https://api.openai.com/v1/threads/thread_JdD0LccyjPsVb4iMk8kd0R0G/messages?limit=100&order=asc".
The conversation went as follows:
Can you tell me about xy?
=> answer was correct and well cited from the file
Can you tell me where i can find this in the brochure?
=> answer was also correct 
Can you tell me exactly on what page?
=> answer was wrong
Why do you tell me it is on page 80 since it is actually on page 140?
=> endless typing symbol with timeout message after while

Both in the frontend and in the backend all became very slow.
In the backend if i would want to access the dashboard of ChatGTP Assistant, of the Assistant Submenu I'll get a timeout error message.

Now, one day later, the answers seem to be more fluid as is the access to backend parts.
My questions :
Can it be that the vector creations was still in process? ( since it is a big file with over 220 pages ) thus slowing down.)
I noticed that the answers are more accurate than just after launch. ( now if i ask for the exact page, they are correct. )
I am using Admin Tools to also generate htaccess file and "protect" the site.
Do I have maybe to configure that to allow connections to openai or similar?

Thanks for your Feedback!

Hello Sébastien and thank you for your purchase of ChatGPT Assistant Pro!

From your description, you did an excellent job following the onboarding process. Bravo!

As you mentioned, the OpenAI API is a separate service, and communicating with it may sometimes fail, as happens with all network-related systems.
A long request may sometimes cause a timeout, or even more likely a moment of increased traffic on OpenAI can take longer than expected.
In ChatGPT Assistant Pro's back-end (e.g. in the dashboard), you can click Options on the top right and go to the Advanced tab.
There you will find an option called CURL Timeout. That is set to 120 seconds. You may increase that if OpenAI answers take more than 2 minutes.
However, I think that was probably a temporary thing rather than something to stress about.

Now, regarding files, I don't think that the LLMs can perform well in identifying the page that the information was found on. I believe the format that files are transformed into (vector stores, embeddings) do not retain the physical information of pages. I may be wrong. If it's smart enough to do so, it's more likely that it just scanned within the text to look for the closest page-like number, rather than anything else. I'm not really sure about this, though.

If it's performing better now, it's probably because you are feeding a simpler question (what page is this on) rather than a more complicated path of thinking (how to xy > where is that found > what page)

AdminTools and htaccess protects inbound traffic. Curl calls to OpenAI are outbound traffic, so you probably don't need to configure anything else.

Let me know if you need any further assistance!