Coming soon: Vision will watch, listen, and read everything you store: spoken lines, faces, logos, objects, on-screen text. Ask in plain language and jump straight to the exact second, in any file, on any of your 90+ storage providers. One search across every connected storage already ships in Performm today. Vision is what comes next.
A full media-understanding stack that reads, indexes, and searches everything inside your files.
Speech-to-text with timestamps, in-frame OCR, object, face and logo detection, scene and shot segmentation, plus summaries and keywords for every file.
Full-text and semantic vector search across every connected store. Ask in natural language and find the moment by meaning, not by filename.
An interactive timeline with visual overlays takes you straight to the exact frame or phrase that matters. Local-first and private by default.
Every result is exposed through the Amove MCP server, so Claude, Cursor, and any agent can query your media content directly.
Vision runs locally for full data sovereignty, built with sovereign AI models, private by default. It works across all 90+ providers Performm already connects, so every store becomes one searchable intelligence layer.
Ships as a native capability in the platform. Every file your team connects becomes instantly searchable.
License the Vision engine on its own and run it against your own media, anywhere.
Vision will turn your video, audio, and image archives into one intelligence layer your team and your agents can search by meaning. Get on the early list and see it first.
Get early access