search

AI-in-Hand / platform

AI Agent Automation Platform: Rapidly prototype, test, and deploy Multi-Agent Systems from your browser.
https://platform.ainhand.com
GNU Affero General Public License v3.0
16 stars 7 forks source link

Using Azure API? #115

Closed robertobalestri closed 1 month ago

robertobalestri commented 1 month ago

Hey, do you think it's possible to use Azure API? I can't find the file where the call request to GPT is done. I'd like to try to modify that to call azure instead of Openai. Thanks a lot

bonk1t commented 1 month ago

Hey @robertobalestri,

The file where the API call to OpenAI is made is located at backend/services/oai_client.py.

Please try following the instructions here: Advanced Usage with Azure OpenAI.

Additionally, you can refer to this related Discord discussion: Discord Link.

For your information, we plan to support the efforts of the Agency-Swarm project to integrate more models and providers in the future. This includes integrating with libraries like the Open Assistant API (see issue #85).

robertobalestri commented 1 month ago

I'm getting crazy in implementing your software, i think i do not have the skills to do it xD I'd ask you something: can you please, in your hosted demo at https://platform.ainhand.com/, add a way to use Azure credentials instead of openai? It shouldn't be difficult: the code just need to call something like this in oai_client: def get_openai_client( user_variable_manager: UserVariableManager | None = None, api_key: str | None = None ) -> openai.OpenAI: if not api_key: if not user_variable_manager: raise ValueError("Either user_variable_manager or api_key must be provided") api_key = user_variable_manager.get_by_key("AZURE_OPENAI_API_KEY") return instructor.patch(openai.AzureOpenAI( api_key=api_key, api_version=user_variable_manager.get_by_key("OPENAI_API_VERSION"), azure_endpoint= user_variable_manager.get_by_key("AZURE_OPENAI_ENDPOINT"), timeout=5, max_retries=5, )) and then where you select small and large model names also there should be configurable with user_variable_manager.get_by_key("AZURE_OPENAI_DEPLOYMENT_NAME_SMALL") and user_variable_manager.get_by_key("AZURE_OPENAI_DEPLOYMENT_NAME_LARGE")

bonk1t commented 1 month ago

@robertobalestri Thank you for the suggestion. I'll have a look today and test the setup.

bonk1t commented 1 month ago

@robertobalestri Now it should be supported as of v0.24.0. I also added the ability to use any custom models in the agent configuration. Feel free to test it and post your feedback on our Discord.