CowCow¶
CowCow is a local-first movement layer for physical AI and robotics datasets: ingest, verify, rank, and sync terabyte-scale multimodal volume from edge to cloud with minimal wasted bytes, time, and retries.
Live surfaces¶
| Surface | URL |
|---|---|
| Web UI | deepubuntu.com/cowcow |
| HTTP API | api.deepubuntu.com |
| OpenAPI | api.deepubuntu.com/v1/openapi.json |
| Demo project | physicalai-demo (reference deployment with sample media routes) |
Start here¶
- Why CowCow: product overview and capabilities
- Install: prebuilt CLI binaries for field laptops and edge devices
- Integration: API routes, auth, and UI wiring patterns
- Movement benchmarks: reproducible sync efficiency numbers
What CowCow does¶
CowCow handles multimodal field data: video, LiDAR, radar, telemetry, calibration files, and MCAP or ROS logs. It content-addresses payloads with BLAKE3, runs local quality checks and priority scoring, resumes interrupted sync, and verifies hashes end to end.
A separate web UI can visualize sync progress and sample previews; sync logic lives in the API and CLI.
Get started¶
Download the CLI from Install, or contact DeepUbuntu for evaluation access and deployment support.