Plato is an AI coding agent that works in a tool window beside your code. It does not hide its work: every step appears as it happens — the files it reads, the commands it runs, the change it proposes — so you can follow the plan as it forms, take the wheel mid-run, or stop at a safe point. Plan, edit, or auto. You hold the switch. Plan reads and reasons; it changes nothing. Edit asks before it writes a file or runs a command, and remembers your answer for the session. Auto works without stopping. Plato cannot change its own mode. When the work needs more room, it asks you. Every step is recorded. Approval cards show the actual diff before a write. Plato keeps undo copies of the files it changed, reads protected files with their values masked, runs the MCP servers you configure, accepts images, and can work Solo, as a Duo with a second model reviewing, or as an Academy with two observers. You hold the switch in all three. You bring the model: Anthropic or any OpenAI-compatible endpoint, including one that runs on your machine. You supply the key and pay the provider directly. A Node.js 22 runtime is bundled, so you install nothing else. Plato sends nothing to Applivity. Prompts and the code it reads go only to the provider you configure. Transcripts, settings, and keys stay on your machine; keys are held by Windows Credential Manager. Policy: https://plato.applivity.com/privacy Plato is pre-release, and this is its first Visual Studio release. It supports Visual Studio 2026 on Windows 11. Support: plato-feedback@applivity.com. |