Build and curate your organization's institutional memory across every source. Subject-level data residency. Source-system access fidelity. Air-gap deployable. Hybrid search, knowledge graph, and supersession on one substrate.
Moʻolelo is a self-hosted federated enterprise knowledge platform for organizations whose institutional knowledge spans multiple source systems — code repositories, documentation wikis, issue trackers, document libraries — and whose compliance posture rules out shipping that knowledge to a third-party cloud.
You install it once into your own infrastructure (Kubernetes today, with optional VM-resident ML workloads). It pulls from your sources, normalizes the records, builds a unified knowledge graph with cross-document reference edges, serves hybrid (lexical + semantic) retrieval over the corpus, and respects every source-system access control end-to-end.
Moʻolelo exposes the same underlying knowledge through complementary surfaces:
All three surfaces share the same chunks, the same access controls, the same audit log, and the same residency boundary.
Moʻolelo is built around a hard invariant: personal data of any individual physically resides only in the regional vault for that individual's residency. The global knowledge graph carries tokens, never PII. Queries federate across regions; the data itself never does. This is the difference between an enterprise knowledge platform you can deploy in a regulated multinational and one you can't. See data residency.
owner/repo#N, Confluence inter-page links, URLs to known sources) and projected as typed edges in NebulaGraph. Stub vertices upgrade in place when the referenced entity is later ingested by its own connector.Operators of regulated multinational organizations — financial services, healthcare, government, defense, regulated manufacturing — who need first-class knowledge intelligence over their institutional corpus but can't or won't push that corpus to a SaaS vendor.
Size doesn't decide it — problem shape does. The same binaries serve a five-person team running everything on a single node and a multinational running half a dozen regions over a petabyte-class corpus.
If you can deploy a Kubernetes Helm chart and you have at least one supported source system, you can run Moʻolelo.
Moʻolelo is a knowledge substrate, not a finished product surface. What ships today is the ingest path, the unified graph + retrieval layer, the supersession engine, and the access-control stack. Forward investment is going into:
Neither is shipped yet. We don't document unshipped features as if they exist — when they land, they get their own pages and a connector framework entry.