Clipto Studio gives ChatGPT, Claude memory of your calls

Clipto is positioning Clipto Studio against a gap it says the big chatbots leave open: assistants can sift through email inboxes and corporate databases, but not the raw video and audio where many decisions are actually made. The company's argument is that this material is stored in heavy, unstructured formats that are expensive for AI to process, so it pre-indexes speech, individual speakers' voices, facial expressions and other on-screen visuals into a structured, searchable dataset before a question is ever asked.

The design choice to run proprietary models locally on Mac and Windows machines, rather than relying solely on cloud processing, is paired with integration through the Model Context Protocol — the same connector Clipto uses to reach ChatGPT, Claude and other assistants. That combination is what lets a user ask about a conversation from two years ago and have the assistant search the metadata of those recordings for an answer.

Clipto frames the launch as a step in a longer push. It recently hired Dr. Mei Han, a former Google research scientist with a doctorate in robotics from Carnegie Mellon University, as chief scientist to lead development of its lightweight multimodal AI architecture. Han described the hard part as connecting information across thousands of recordings so AI can understand context, rather than merely recognizing speech or identifying what is on screen.