Run AI on Your Device: Getting Started with Microsoft Foundry Local

What if you could ship AI features inside your app with no Azure subscription, no per-token charges, and no data leaving the device?

In this session, we'll explore 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁 𝗙𝗼𝘂𝗻𝗱𝗿𝘆 𝗟𝗼𝗰𝗮𝗹 — an open-source, ~20 MB runtime that embeds directly in your application and runs models like Phi-4, Qwen 2.5, and DeepSeek-R1 locally on Windows, macOS, and Linux.

𝘞𝘦'𝘭𝘭 𝘤𝘰𝘷𝘦𝘳 𝘵𝘩𝘦 𝘞𝘩𝘢𝘵, 𝘞𝘩𝘺, 𝘢𝘯𝘥 𝘏𝘰𝘸 𝘪𝘯 𝘵𝘩𝘪𝘴 𝘴𝘦𝘴𝘴𝘪𝘰𝘯:
• What it is: architecture, the curated model catalog, and how it differs from Azure AI Foundry
• Why it matters: privacy-first AI, offline scenarios, zero inference cost, and edge/desktop use cases
• How to build with it: 4 live Python demos — streaming chat, multi-turn conversations, tool-calling agents, and OpenAI SDK compatibility (drop-in replacement, no code rewrite)
You'll see everything run live, locally, on a real Windows 11 machine — no cloud calls, no API keys, no surprises.

𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆𝘀:
• Understand when to use on-device AI vs cloud AI (and when to combine both)
• Know how to set up Foundry Local on CPU-only and GPU machines
• Leave with working Python demo code you can adapt immediately