MagenticLite
MagenticLite is an agentic application from Microsoft Research AI Frontiers, the next generation of Magentic‑UI, designed to make it dramatically easier to drive an AI agent across both your browser and your local file system in a single workflow. The redesigned app pairs a refreshed chat and browser view with an agent harness rebuilt for small models — actively curating context, planning step by step, and delegating browser tasks to specialized sub‑agents. Transparency is built in: you see what the agent is reasoning about, you can take direct control at any moment, and critical actions like logins or irreversible submissions pause for explicit approval. In short, it’s an agent stack you can run, inspect, and trust.
MagenticLite is built for developers, researchers, and power users who want capable agents that run closer to them — without the heavyweight compute bill of frontier models. It’s also the canvas for a broader research bet from Microsoft Research AI Frontiers: that agentic capability lives in tool orchestration and action, not raw model knowledge, so a co‑designed app, harness, and pair of small models can deliver strong performance at a fraction of the cost. Browser sessions and code execution run inside Quicksand, the project’s open‑source QEMU sandbox, so your data stays on your machine.