Not launched yet · building in the open
Data built to move fast, not sit still.
One API. No server to babysit.
Shard Mesh breaks your data into small, fast-moving chunks and spreads them across a peer-to-peer network instead of one slow database. You call one API — we handle the rest. Other tenants can't see your data, not even by accident.
No single database to bottleneck.
Low volume. Only when something real ships. Unsubscribe any time.
The pitch, fast
The usual "realtime" stack
- Data sits on one database until it doesn't fit anymore
- Scaling means a bigger, more expensive box
- Tenant isolation is a filter you hope catches everything
Shard Mesh
- Data is chunked and spread across a mesh built for fast transfer — no single server to outgrow
- One API and a token. No infrastructure for you to run
- Tenant isolation enforced on every row, every request — not a best-effort filter
Fast because nothing sits in one place. Safe because nothing crosses tenant lines.
What's running under the hood
You never touch any of this directly — it's what the API is built on. Each piece does one job, has strong opinions about it, and isn't shy about what it doesn't do yet.
Realtime pub/sub
Topic streaming over authenticated links. On by default. Never opens a storage handle.
built · off by defaultCRDT shard storage
LWW-maps and OR-sets, hash-ring placement, anti-entropy. Durability you opt into per node.
built · partialSQL-shaped layer
Tables and secondary indexes over the shard store. A Go API today; a real parser later.
builtMesh transport & trust
Mutual TLS 1.3 between peers that both carry a certificate signed by your mesh root.
builtmeshctl
Stand up a mesh, mint identities, issue certs, read a node's real status. One binary.
soonThe capacity brain
One server that watches demand, predicts need, and assigns roles. Not built yet.
Questions people actually ask
What is Shard Mesh?
An API for realtime data and storage. You call one endpoint; underneath, your data is broken into small chunks, packed in a format built for fast transfer, and spread across a peer-to-peer mesh of nodes instead of sitting on one database. No server for you to run or manage.
Is this blockchain or crypto?
No. "Peer-to-peer" here is purely a data-movement architecture — nodes replicate chunks of data between themselves for speed and redundancy. No token, no ledger, no consensus mechanism, and you never talk to the mesh directly; everything goes through one API.
What does it cost?
Free tier for realtime and light storage. Paid plans ($29–149/mo) unlock regional edge acceleration, higher limits, and durable storage. The mesh itself runs on cheap, commodity nodes — that's how prices stay low — but you never provision or manage any of that yourself.
Who owns the data, and can other customers see it?
You own it. Every row is tagged to your account and checked on every single read — a request for data that exists but isn't yours returns the exact same response as data that doesn't exist at all, so another tenant's data can't even be detected, let alone read.
When does it launch?
Hasn't yet — it's being built in the open. The gateway API, realtime layer, mesh storage, and transport security are built already. Join the waitlist for updates, or add a request to vote on what ships next.
Bored? Tell us what to build.
We're deciding what lands next, and the deciding vote is yours. Add a request on the waitlist page, upvote one that's already there, watch the list move. That's the whole roadmap process — no committee, no quarterly planning doc.