OriginChain
solution · fraud detection

Detect rings the moment they form.

Three engines that usually live in three systems - streaming, graph, rule evaluation - composed on one substrate. Zero replication lag between them.

three engines you usually need
01
Streaming

React to every write the instant it lands. /watch over the transaction prefix is a one-line subscription.

02
Graph

Find rings, detect collusion. PageRank ranks influence; triangle enumeration finds 3-clique mules.

03
Rules

SQL predicates over customer + transaction tables. Predicate pushdown keeps rule evaluation fast.

why one substrate matters

Fraudsters exploit the gap between systems.

Split the stack into Kafka + Neo4j + Postgres and your graph view is 30 seconds to 5 minutes behind reality. A ring spinning up in that window is invisible to the detector that has the right algorithm.

WAL-native /watch + graph queries against the same store means zero lag. The detector and the data are the same surface.

subscribe
curl -N \
  -H 'Authorization: Bearer ...' \
  'https://<tenant>.db.originchain.ai/v1/tenants/:t/watch?prefix=transactions/'
triangle query (collusion)
POST /v1/tenants/:t/graph/triangles
{
  "schema":   "transfers",
  "min_amount": 5000,
  "since":   "PT2H"
}
deferred

See /watch and the graph surface.

/watch details → Graph Try it