Vivu
Vivu is an AI search layer for video libraries that indexes what is said, shown, and written on screen, letting teams find moments by meaning rather than filename. It connects to AI agents like Claude and ChatGPT through a remote MCP server, and offers an on-device model for real-time video detection on edge hardware like phones and robots.
- Artificial Intelligence
- Enterprise Software
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Companies accumulate large video libraries across marketing, sales, training, and operations, but these archives remain effectively unsearchable—findable only by filename, manual tags, or the memory of the person who shot them. Teams waste hours scrubbing through footage to locate a specific moment, often re-shooting content they already own because the original cannot be found.
Solution
Vivu provides an AI search layer that reads the actual content of videos—what is said, what is shown on screen, who is speaking, and the text displayed—and connects these signals into a structured, queryable index called the VCG (Video Content Graph). Users search their entire library in plain language and receive candidate moments as time ranges with source footage they can open and verify. The platform runs a remote MCP server that lets AI agents like Claude and ChatGPT query the same library, bringing video content directly into existing workflows. Each recording is indexed once at ingest, making it cost-effective to search at library scale, and the index compounds with every video added. For robotics teams, Vivu's optimized on-device model brings real-time video detection and indexing to edge devices, helping improve how vision-language models recognize objects and understand relationships in the physical world.
Target Audience
Vivu targets marketing teams, sales and enablement organizations, training departments, event teams, AI and product teams, robotics and remote operations, customer support, security surveillance review teams, and post-production/broadcast archive teams that need to search large, unstructured video libraries by meaning.
Features
- Plain-language semantic search across video content, indexing speech, visuals, on-screen text, and speaker identity
- Source-backed results returned as candidate time ranges with timecodes that link to the original footage
- Remote MCP server connector enabling AI agents (Claude, ChatGPT) to search the video library directly from chat
- Single-pass indexing at ingest that avoids reprocessing on each query, keeping costs low at library scale
- Browser-based upload and search requiring no engineering setup or file reorganization
- On-device optimized model for real-time video detection and indexing on phones, Raspberry Pi systems, and robotics hardware
- Workspace-level privacy controls with media deletion removing content from Vivu's cloud