Land every dataset and every event in one open, Apache-Iceberg lakehouse — then ask it questions in plain English. Distributed SQL, event projections, and hard per-tenant isolation, on your own infrastructure.
Open storage, a smart catalog, and enforcement you can prove — from ingestion to query.
Durable Apache Iceberg tables on your object store — readable by Spark, Trino, and DuckDB. No lock-in, no proprietary format.
A governed AI catalog assistant: ask a question in plain English, get SQL, charts, dashboards, and pipelines — scoped to what you're allowed to see.
Serverless queries for exploration and a Spark/Livy engine for scale — the same catalog, the right compute for the job.
Land raw CloudEvents, then project them into typed views automatically — including W3C trace context for distributed-transaction search (OTEL).
Every table, row, and read is scoped to its tenant by construction — cross-tenant access is structurally impossible, enforced with OpenFGA ReBAC.
Self-hosted and air-gap capable, with masking, access requests, and an auditable trail for every governed action.
Four steps from an empty warehouse to a catalog anyone can query — safely.
Stream documents, tables, and CloudEvents into Iceberg — batch or continuous.
Turn opaque event payloads into typed, browsable views with one click.
Grant, mask, and request access per catalog, schema, table, and column.
Query in SQL or ask Mnemo in plain English — always within your permissions.
Access is enforced at the edge and in the data plane — not bolted on.
OpenFGA relationships decide who can read, mask, or manage each object — down to the column.
Ingests, queries, grants, and deletes are recorded as an immutable, distributed-traced trail.
Bring your own LLM provider and credentials; the catalog assistant runs within your governance boundary.
The live demo is pre-loaded with documents, sensor data, and trace-correlated events — no login required.