/
Profound AI FAQs

Profound AI FAQs

If you have a question about your Profound AI instance that is not answered here, please reach out to our Support Team.

 

  1. What is the minimum version of Profound UI that will work with Profound AI?

    1. The minimum Profound UI version that will work with Profound AI is 6.24.0.

  2. Which tiers of Profound AI have logging included?

    1. Logging is only enabled for Business and Enterprise tiers of Profound AI. This also means that the ai_chat.log is only available for these tiers as well, and it is not available in the Standard tier.

  3. Can I use the AI Agent on a different domain?

    1. Yes! Although, there are some important details that should be kept in mind:

      1. If you are using the Agent on a different domain and a different port, you will need to enable CORS for the Agent in the following two places:

        1. The Deploy Agent dialog.

        2. The enableCORS configuration setting.

      2. If one domain is HTTPS, the other domain must also use HTTPS.

  4. Do I need to have a paid account/version of the Large Language Model (e.g., OpenAI, etc.) in order to use it with Profound AI?

    1. Yes, it will usually be required that you have a paid version of the Large Language Model you are trying to use. This is because most of the Large Language Models will only work with Profound AI if you have a paid version.

  5. My Agent isn’t working in the IDE. What should I do/check?

    1. If your Agent isn’t working in the IDE, we recommend that you check your instance version.

      1. If you are running an instance of Profound AI that is older than 7.13.0, then you will need to upgrade to 7.13.0 or newer. This is because a bug was identified in older versions that related to this, and it was fixed in Profound AI version 7.13.0.

  6. I get an error when uploading Knowledge Documents for my Agent.

    1. If you receive an error when uploading Knowledge Documents, make sure that you are using a model that utilizes the “assistants” endpoint/API.

      1. This functionality does not work with the “completion” endpoint/API.

    2. Currently, OpenAI models use “assistants” endpoints.

    3. The assistants endpoint is not yet available via Azure OpenAI (as of mid-2025).