Chatting with OpenAI (gpt-4o) using 20 tools.

What is this?

This is a prototype of OpenAI function calling with GPT 4 Turbo. The code behind this prototype is rather simple. The model is provided a list of retrieval functions, and when prompted, is allowed to recursively call these functions until there is sufficient context to answer the user's question.

How to use this prototype:

  • Messages are stored in your browser. Clear the conversation by refreshing or clicking the clear button
  • Use the input below to start a conversation with Copilot
  • On the right, you'll see the functions called by the model. Click on a function to reveal the raw response body and function schema
  • Click on the cog icon in the top left to modify settings

Try it

How can you help me use GitHub?
Summarize the last 3 changes to vercel/ai
Search for all important files related to the Dialog component on primer/react, then retrieve the last commit for each file, then summarize the diff associated with that commit
Summarize comments on the last 3 pull requests to primer/react
Search hackernews and summarize sentiment for vercel and their role in AI development. Then look at the first 3 pages of issues in verce/ai and help me prioritize 3 new milestones.
Summarize the intent behind pr #4035 in primer/react. If the PR mentions any related issues, fetch those issues to inform your answer.
Find the most recent issue in primer/react, then use code search to suggest files related to that issue, then suggest changes to those files that help me get started with the issue you selected.
Functions, tools, and prompts will appear here.
Open settings to change function calling strategy. Tools are run in parallel whereas functions run sequentially.