One endpoint. Four query shapes. One engine. Rows, vector
embeddings, full-text postings and graph edges commit on the
log together - every write atomically visible to every shape,
including the natural-language /ask on top.
Single-tenant. Region-isolated. Sub-100 ms in-region.
$ curl https://oc-thunder.ap-south-1.originchain.ai/v1/ask \
-H "Authorization: Bearer $OC_TOKEN" \
-d '{"q": "last 7 closes for NIFTY"}'{
"plan_id": "tpl_nifty_close_lastN",
"rows": [
{ "date": "2026-04-15", "close": 22456.30 },
{ "date": "2026-04-16", "close": 22512.85 },
{ "date": "2026-04-17", "close": 22489.10 },
{ "date": "2026-04-18", "close": 22534.75 },
{ "date": "2026-04-21", "close": 22601.40 }
],
"meta": { "latency_ms": 42, "cache_hit": true,
"region": "ap-south-1" }
}
Working with NVIDIA on GPU-accelerated vector search, managed embedding hosting, and managed LLM serving. Learn more →
Most modern AI stacks fan a single user action across four databases - a relational store, a vector index, a search engine, and a graph. Each has its own bearer, its own SDK, its own billing line, and its own way of being out of sync at 3 am. OriginChain replaces all four with one managed database. One write lands every shape atomically.
Retire four contracts. Onboard one. Stop coordinating four maintenance windows for a single feature release.
The row, its embedding, its full-text postings, and its graph edges commit on the log together. No reconciliation jobs. No "the search index is 8 hours behind" disclaimers.
One SDK, one auth model, one observability surface. New AI features land without a Debezium pipeline + four retry queues + a 3 am reconciliation cron.
SQL, vector, full-text, graph, and natural language all run against the same managed database - single tenant, region-isolated, with replication and point-in-time recovery you don't have to wire up.
SQL · Vector · Full-text · Graph · Natural language, all on the same store and the same bearer. Write your data once, query it four ways without bolting on a separate engine.
p99 under 8 ms for typed SQL. HNSW vector top-k at 100k vectors: recall@10 = 0.96 with p99 109 ms in default high_recall mode, or p99 37 ms in fast mode (recall 0.69). Warm natural-language queries return in under 50 ms thanks to plan caching.
Your own compute, your own volume, your own bearer - locked to the region you pick. Zero shared resources, zero noisy neighbours, zero cross-tenant blast radius.
Provisioning, TLS, backups, replication, and observability run for you. You ship product features; we run the database underneath them.
POST a sentence to /ask and get rows back. Plan-cached natural-language queries execute in-region with sub-50 ms warm latency and a stable schema-aware response shape.
RPO=0 on paid tiers and ~25 second failover keep your application online through hardware loss. The WAL is continuously archived to encrypted object storage, so you can restore to any timestamp.
Every customer gets a dedicated database instance in their chosen region. No shared compute. No shared storage. No cross-tenant query path. The architecture is the compliance answer - not a checkbox bolted on top.
GDPR, DPDP, and sector-specific data-residency rules are satisfied by where the bytes live and who can touch them - and you hand both answers to an auditor in writing.
One dedicated instance per customer, not a shared cluster. Noisy neighbours cannot exist by construction.
Frankfurt, Paris, Stockholm - pick where the bytes live. EU-resident data never leaves the EU.
Each tenant's bearer scopes to its own ULID. A leaked credential cannot reach another customer's instance.
DPA, sub-processor list, deletion timelines all in writing. Self-service account deletion with a 30-day grace window.
Concrete p99 numbers, not averages - the ceiling your app can plan around for SLAs, agent loops, and user-facing reads on a managed database.
End-to-end p99 in-region for typed queries against your managed database.
After the first ask of a shape, the plan is durably cached and every repeat skips compile.
Default high_recall mode hits recall@10 = 0.96 at 100k vectors. Fast mode runs p99 37 ms at recall 0.69 when latency dominates.
First ask of a brand-new shape from another continent - including the round trip.
p99 measured at the API edge · in-region unless noted · vector topk on managed HNSW with f32 SIMD distance kernels, default high_recall mode (recall@10 = 0.96 at 100k)
Open a connection to /watch
and receive row-level inserts, updates, and deletes the moment they
commit. Standard server-sent events, ordered, exactly-once,
resumable from any cursor - no client SDK required.
Power live dashboards without a polling loop. Drive agent loops on fresh state. Invalidate caches the instant a row changes. The same managed database serves your queries and pushes your changes - no separate change-data pipeline to operate.
curl -N https://oc-thunder.ap-south-1.originchain.ai/v1/tenants/acme/watch \
-H "Authorization: Bearer $OC_TOKEN" \
-H "Accept: text/event-stream" \
-d '{"table": "orders", "where": "status = \'pending\'"}'event: row.insert
data: { "id": 48201, "customer": "Aether Labs", "total": 1842.00,
"status": "pending", "ts": "2026-05-01T09:14:22.018Z" }
event: row.update
data: { "id": 48198, "status": "pending → paid",
"ts": "2026-05-01T09:14:22.214Z" }
event: row.insert
data: { "id": 48202, "customer": "Northwind", "total": 612.50,
"status": "pending", "ts": "2026-05-01T09:14:22.881Z" }
event: row.delete
data: { "id": 48180, "reason": "cancelled",
"ts": "2026-05-01T09:14:23.402Z" }
// ... the connection stays open. Every change to a matching row
// arrives within milliseconds, in order, exactly once.Push row-level changes straight to the browser. The dashboard stays in sync with the database to the millisecond, on any tier, in any region.
Wake your agent when the world changes, not on a timer. Subscribe to the rows that matter and react the instant they commit.
Bust the right cache key the moment a row changes. No TTL guesses, no race windows, no stampede when many keys expire at once.
OriginChain is the AI-native database for teams that need SQL, vector, full-text, graph, and natural-language queries on one managed endpoint. Single tenant, region-isolated, fully hosted - provision in under two minutes, query in milliseconds.
Pick a region, pick a tier, and we provision a dedicated single-tenant instance in roughly ninety seconds. The first query you send is the first query we'll show you how to write - in English.