Build agents with visual intelligence
Your agent can now watch video. Ask it questions, find the exact moment, and get back a clip you can stream.
1,000+ developers and agents use VideoDB every week. 300+ TB of video processed.
Any MCP clienthttps://mcp.videodb.io/mcp
Teams using VideoDB every day
- 100,000 hourschecked and cut into training samples in under a week for one of the largest video data providersRead the case study
- 3 weeksfrom recordings to a first computer-use agent for a Big Four accounting firm’s capability centersRead the case study
- 2×the next pipeline’s score on a 100-hour visual moment retrieval benchmark, when the clip must line up with the momentSee the benchmark
Add VideoDB to the agent you already use
It takes about a minute. Then paste a link or upload a file and ask.
Claude Code
Three commands in the terminal.
/plugin marketplace add video-db/skills
/plugin install videodb@videodb-skills
/videodb setupCursor, Codex or any MCP client
Add one URL as a remote server.
https://mcp.videodb.io/mcpWhat agents do with VideoDB
Kinds of agents built on VideoDB, a typical request to each, and what comes back.
-
Archive agents
“Cut three Shorts from this podcast.”
What comes back: search_video finds 3 moments; reframe_video makes vertical clips, 9:16.
-
Computer-use agents
“What did the agent click before the error?”
What comes back: search_video finds Submit invoice, at 14:02:11.
-
Physical agents
“Where did you last see the blue bottle?”
What comes back: search_rtstream finds it on the kitchen counter, at 09:41:07.
-
Research agents
“Two arms picking up objects in a kitchen.”
What comes back: dataset_search finds episodes from 9 datasets; export_result_set writes LeRobot files.
-
Fleet agents
“A cyclist cutting across at dusk.”
What comes back: search_collection finds clips, with drive, time and camera.
Memory for agents and physical devices
Everything a camera, a screen or a file shows goes into an index as it arrives. A task gets back only the frames it needs.
Read how a robot remembersFirst on a 100‑hour visual moment retrieval benchmark
VideoDB finds a matching clip for more queries than any other pipeline tested, on an open dataset with 11,369 queries.
Read the benchmarkQueries with a matching clip in the top 10
- VideoDB
- AWS Nova
- Twelve Labs Marengo 3.5
- SigLIP2
- Cosmos-Embed
- Qwen3-VL Embedding
- InternVideo2
- PE-AV
From your files to your agent
Storage, search and streaming for video, in one system.
-
About half a second from a question to the frames.
Read about delivery -
Any model describes every scene. Search runs on what it writes.
Read about indexes -
Buckets, cameras and screens, read in place.
Read about storage
Open-source agents to start from
5,400+ stars on GitHub. Each repo runs on VideoDB and is free to fork.
VideoDB on GitHubThe most-starred VideoDB repositories on GitHub. Each opens in a new tab.
Where VideoDB runs, and how it keeps your data safe
Runs on any cloud
AWS, Google Cloud and Azure, or your own cloud.
AWSGoogle CloudAzure
Security and compliance
SOC 2 Type II, GDPR and ISO 27001 compliant.
What builders say
-
Oliver CameronFounder of Odyssey, ex-Voyage and Cruise
For a long time, I have been thinking what you are building needs to exist! Glad you are building it!
-
Tom MasonFormer CTO, Stability AI
VideoDB team, nice work! Would love to chat!
-
Johnny BoufarhatFounder of Hopin
VideoDB is cool!
Recent research
Most of what we publish is measured on public benchmarks, with the numbers included.
All research-
VideoDB leads a 100-hour visual moment retrieval benchmark VideoDB against AWS Nova, Twelve Labs and five open embedding models, on an open dataset with open queries. Learn more
-
How a robot remembers How a robot keeps an index of everything it has seen, and pulls back the few seconds a task needs. Learn more
-
Search over the Visual World Persistent visual memory, layered indexes and source-grounded evidence for agents over cameras, screens and archives. Learn more
Try it on your own video
$20 of free credits and no credit card. Add it to Claude or ChatGPT, then paste a link and ask.


