A database · not a sync layer
One database engine. Embedded on a phone, or clustered across a datacenter. Offline, elastic, and versioned — three things your team stops worrying about.
What disappears
Every instance holds the full history. Replication and merge aren't modules stacked above the database — they're how the engine works. Disconnect, keep writing, reconnect, and only what changed crosses the wire. Users never lose a keystroke.
Peers discover each other and data rebalances itself. No replica configs, no connection poolers, no migration weekend. Adding capacity is boring — which is the highest compliment you can pay infrastructure.
Every write is permanent and addressable. Branch your data like code. Merge experiments. Roll back to any moment. A complete audit trail, woven in — not added after compliance asked.
Stream the write log to S3 (or any object store). Servers themselves become ephemeral — spin them up, scale them out, tear them down without migrating state. Durability lives in cloud storage; servers only need speed.
Where we are
If you're fighting a sync layer, a sharding project, or an audit bolt-on — we'd like to hear about it. Twenty minutes is enough to decide if it's interesting.