Reclaim the power of creation.
PeerWeave is a sovereign peer-to-peer AI platform that unifies development, knowledge, and collaboration into a secure, intelligent mesh. We’re building the substrate for people who want their tools—and their data—to belong to them again.
Why the world needs this
Modern stacks lean on centralized platforms for code, CI/CD, AI runtimes, and collaboration. They’re incredible—until they fragment workflows, compromise privacy, or price out experimentation.
We read every alternative we could find (GitHub, Radicle, Holochain, graph DBs, agent orchestration frameworks) and none of them combine local-first workspaces, peer-to-peer sync, semantic graph lineage, and agent-native services in one place.
What we’re building instead
- • Local CRDT spaces that sync peer-to-peer.
- • A semantic graph of files, functions, agents, services, and history.
- • Ports that turn any device into a micro-cloud.
- • Mirrors + anchoring to share what you choose, when you choose.
Think of PeerWeave as the OS layer under your IDE and agents—not a replacement for them.
Principles we won’t bend
These pillars come straight from the Differentiators doc. They guide every product decision.
Local-first
Spaces live on your hardware. Sync happens peer-to-peer over CRDTs. Cloud is optional.
Agent-native
Humans and many AIs collaborate in the same workspace, sharing context through the graph.
Sovereign identity
DID-style keys, capability tokens, and end-to-end encryption keep ownership with you.
Peer-to-peer fabric
No central servers, no gatekeepers. Devices discover each other and sync directly.
Who it’s for
PeerWeave feels familiar if you live in IDEs, experiment with agents, or build distributed systems. But we keep the copy in human language so anyone can get it.
Indie developers
Ship faster with offline-first sync, context-rich agents, and publishing you control.
Teams
Gitless collaboration, shared agents, encrypted workspaces, and zero lock-in.
AI builders
A multi-agent runtime with trace logs, permissions, and a shared semantic graph.
Researchers & creators
Host experiments locally, collaborate privately, then mirror only what you want public.
Technical foundation
Under the hood we combine CRDT-based sync, libp2p networking, semantic graph indexing, and WASM-sandboxed agents. Every component defaults to local-first behavior with optional cloud mirrors.
Identity is DID-driven, collaboration uses MLS group encryption, and mirrors can be anchored to EVRMORE for provenance. It’s opinionated enough for a great out-of-the-box experience but flexible for custom setups.
Read the technical architecture →Join the movement
PeerWeave is the connective tissue between IDEs, agents, services, and people. Help us prove that local-first, peer-native tooling can feel better than anything running in someone else’s cloud.