Embedding a GPT in website

Hi Chris,

My wish is to run a GPT on our own research database with the purpose that people can interact with the wisdom, same as you interact with ChatGPT.

The current GPT we made in OpenAI is not shareable for the public without a OpenAi Plus account, could I use this solution to get the same outcome?

That when people go to our website, this is where they start (see screenshot).

Looking forward to hear.

Asked with love,

1 Like

You should try GitHub - SamurAIGPT/Open-Custom-GPT: Create Custom GPT and add/embed on your site using Assistants api which allows you to build a Custom GPT and embed on your site and it’s open-source


This looks like a decent solution to me.

chatbase .co

Off course this is possible.

Here: https://platform.openai.com/docs/assistants/tools/uploading-files-for-retrieval

Chunking, vector store and retrieval are taken care of. You can build a front end just like playground, and fit it all together. If you need help with this, lets connect.

You can use Actin by MOBJAI. It helps embed your gpts to your website

Hi all, thanks for the lively discussion!

I found the solution:

Meow Apps AI Engine chatbot now has Assistants integration (in their pro plan so 50 bucks a year).

So simply select your assistant and embed the chatbot via wordpress or other means.

1 Like

A quick question as I am a little confused. Is the Open Ai assistant the same as custom gpt? So can it be also used with own knowledge from a file like Custom GPT?

@jan.breidenbach Yes it is the same indeed. The setup process doesnt have the “conversational” setup process as the GPTs have but if you already have a GPT you can copy the “instruction” text and paste it in you assistant.

Yes it is different indeed.

An assistant cannot perform actions that access a developer’s authenticated API directly, there is no internet access provided by OpenAI.

That lack of internet access also includes no “Browse with Bing” function.

Answering with file retrieval and the provided internal function for letting the AI explore uploaded file data can become expensive as it uses iterative calling with the language tokens billed at the account holder’s expense.

It’s going to require a bit of work since I had some issues at first, but that’s only because I was learning as I went along.

But essentially you substitute all user inputs on the Python side by routing the user inputs from the HTML page, and vice versa for the outputs.

Most of the Python code remains functionally the same as the examples.