Event banner
Event details
Is there an SDK or set of API's to have a GPT-4o audio agent join a Teams call? Could multiple agents join a teams call? eg: a Project manager agent, a QA agent, and a few human developers all in the same Teams call? Is this possible today? If so, which SDKs/APIs would enable this use case?
Travis_Wilson_MSFT
Microsoft
MicrosoftWe don't yet have higher-level abstractions for Teams specifically, but between OpenAI and Microsoft we've started some client library coverage to expose the new capabilities of the gpt-4o-realtime-preview model and the /realtime API:
- The OpenAI .NET SDK (https://github.com/openai/openai-dotnet) (as well as the AOAI companion library, Azure.AI.OpenAI) already has early support for a client integrated into 2.1.0-beta.1
- Python has an early standalone client we're iterating on at aoai-realtime-audio-sdk/python at main · Azure-Samples/aoai-realtime-audio-sdk (github.com)
- JavaScript has an early standalone library: openai/openai-realtime-api-beta: Node.js + JavaScript reference client for the Realtime API (beta) (github.com) and we also one at aoai-realtime-audio-sdk/javascript at main · Azure-Samples/aoai-realtime-audio-sdk (github.com)
We've already seen developers prototype applications with multiple agents talking to people (and each other!) using the /realtime capabilities and the results are very cool. Very possible with the tools we have today!
- Excellent, thanks for the reply, I'll take a look.