Multimodel Multimodal and Multiagent innovation with Azure AI
summary
Live demos show GPT‑4o handling vision tasks (no speech yet) inside Azure AI Studio alongside the new Phi‑3 family (Mini, Small, Medium, Vision) and 1,600 models in a catalog; a laptop runs a small model via ONNX Runtime Web. A video translator preview renders speech into other languages in the original voice. Agents autonomously write and execute Python, GitHub Copilot Workspace generates C# clients and unit tests, and a student tutoring app wins a 1,000‑registrant hackathon; enterprise deploys cite 53,000 live customers.
AI-written summary · about
chapters (14)
- 0:00Multimodal AI innovation
- 3:07GPT-4O vision capabilities
- 6:38Video translation demo
- 8:04Phi-3 small models
- 10:14Phi-3 in browser
- 12:24Responsible AI in Azure
- 15:33Safety: content filtering
- 20:11Contact center AI apps
- 21:15AI agents for spreadsheets
- 24:10GitHub Copilot Workspace
- 26:20Unity Muse AI tools
- 36:35Muse Animate text-to-anim
- 40:52Hackathon winners
- 43:43Wrap-up
related talks
51mThe Next Frontier: Azure OpenAI Models and the Arrival of GPT-4
25mMultimodal Conversational Interfaces with GPT and Vision AI
59mLearn Live: Introduction to Azure OpenAI Service
59mBuild Next-Gen AI Apps with .NET and Azure
31mAzure AI Innovations at Microsoft
48mGetting started with generative AI using Azure OpenAI Service
48mUsing Azure AI Foundry to manage all your Large Language Models
01h 34mHands-on AI workshop: Multimodal agents