OriginChain
01 · the substrate

One database. Every shape. One write.

We replaced Postgres + a vector DB + a graph DB + a search engine + a watcher with one substrate. Every write across every shape commits in one WAL frame - by construction, not by best-effort.

01 Atomic cross-shape writes

One WAL frame, every shape. Either every shape sees the row, or none of them do.

  • ·single-writer contract
  • ·byte-level WAL invariant
  • ·cutover atomic at the substrate level
02 Hash-keyed k/v

No relational engine, no second physical store. Shapes are key shapes on one substrate.

  • ·O(1) point lookups
  • ·key-prefix scans
  • ·secondary indexes are first-class
03 Snapshot-bootstrap HA

Failover restores full state in seconds, not hours.

  • ·2026-04-30 drill passed
  • ·frame-stream replication
  • ·leader-election via durable lease
04 Dedicated per-tenant compute

Your data plane is yours alone. No shared engine, no noisy neighbour.

  • ·region-isolated
  • ·single-tenant
  • ·no shared multi-tenant engine
by design
6
query shapes shipped
9
graph algorithms
4 × 4
vector metrics × index variants
1
WAL frame per cross-shape write
0
cross-store sync jobs
design choice

Why a single substrate, not a multi-engine stack.

We picked a single hash-keyed k/v substrate so cross-shape atomicity is free. Adding a separate columnar store would have given us better analytical scans at the cost of atomicity - and we judged atomicity the moat.

The trade-off shows up in workloads. Multi-TB historical scans aren't our sweet spot; warehouses do those better. Atomic cross-shape writes on fresh, operational data are.

why this matters for AI
  • RAG. Sparse + dense + the original row, all in one atomic upsert. The retrieval can't return a vector pointing at a row that no longer matches.
  • Agent memory. JSON-shaped values + embeddings + lineage, transactional. The agent's view never lies about what was just written.
  • Streaming personalization. /watch sees every write the instant it lands - including the new vector.
what we didn't do
refused
Multi-engine architecture

A separate store per shape would have meant cross-store coordination on every write. We don't do it.

refused
Eventual consistency

Single-writer is the contract. Read-your-writes is immediate. No replication-lag explainers in your runbook.

refused
Self-host edition

Managed only. Operating the substrate at the SLA we promise is our job, not yours.

go deeper

The full architecture, with every shape drawn out.