How can i build my chatgpt by speacial knowledge database

I want to establish a ChatGPT for a specific domain based on a particular knowledge base. How can I do that?


That’s a good question! Wiould love to see smart people answer it

Assistants API with fine-tuning + RAG for the most critical documents. Check it out here.

Hey I have a lot of experience doing this. Is this a public knowledge base where you can send a URL? Or is it a private knowledge base?

What format and roughly how large?

LMK and I’ll recommend best path.


(at risk of hijacking isozhj’s question…) we have a similar need, and our domain-specific knowledge base is technical in nature and sits for example across 10’s of word and powerpoint documents. We are creating a custom GPT to help with report and proposal generation on the basis of that domain of knowledge.

I assume that we would just upload all those files into the “Knowledge” section, under “Configure” section for the GPT Builder?

How do we make sure this information that we upload stays private?

Thanks in advance for your help!

Hey @joe-z , in general it’s best to change the files to the most size efficient format (typically grouping the docs together into one text file is a good way to go).

Then yes you can upload the docs directly to Knowledge.

As for Privacy, any document that is uploaded can be easily downloaded in its entirety.

A better approach is to use a GPT Management system to manage your Knowledge and it takes care of the step of grouping documents and minimizing size.

You then would have the GPT access the knowledge via and Action (rather than directly uploading).

Note you can restrict who accessed to GPT to just whomever you send the link to or if you are on GPT Enterprise, to your organization.

Feel free to DM if I can further advise.


I have a follow up question, I try to upload a word and a text file to the knowledge section of the custom GPT but it keeps disappear after saving the changes (no error appears), and accordingly the GPT doesn’t use it (so it’s not integrated into it somehow). Do you have an idea what should I do? is it bug?

Hello, I would like to know how many PDF files can I add to the knowledge section. I add more than ten files and when I add the eleventh file I get this error message “Error saving draft”.

Is there a limited number of uploads?
Is there a way to upload more files?

  1. . Convert all text files (.pdf, .docx, etc.) to text files. Ideally format as Markdown (.md). .txt is fine too.
  2. Convert all table files (.csv, Google Sheet, etc.) to .xlsx as ChatGPT works especially well with Excel files.
  3. Per GPT Limits: 10 files
  4. Per File Limits: 512MB (20MB for image files, no limit for .xlsx), 2M tokens
  5. Per User Limits: 10 GB. Per Organization Limits: 100 GB.

Thanks @cass .

I do not (yet) have access to GPT Enterprise.

…BUT IF I keep the custom GPT private (i.e., restricted to only those whom I send link to), will the uploaded Knowledge files be still accessible somehow by the public (i.e., by becoming “part of the (general GPT 4) model / ether”)?

My understanding of GPT Enterprise is that they specifically state they will NOT use your data for training their models.

1 Like