Architecture

AthenaDB is a single engine, built from the ground up. There is no embedded third-party database — the storage format and cryptography are ours end to end, which is what makes them auditable.

The engine

At the core is an encrypted object store with its own indexing and query planner. Everything above — the retrieval surface, the domain packs, the sync layer — is built on that one engine.

One retrieval surface

AthenaDB unifies four retrieval modes behind a single hybrid-RAG query API:

• Keyword / full-text — classic inverted-index search.
• Vector — semantic similarity over embeddings.
• Graph — relationships between entities.
• Timeline — time-ordered retrieval.

A single query can blend these, rather than forcing you to choose one index type. Details: Retrieval & embeddings.

Form factors

The same engine appears in several shapes (Early Access status varies):

• Lite — embeddable single-file mode for desktop and edge (In progress).
• Server — networked instance for multiple clients (In progress).
• Sync — instance-to-instance sync over the Trust Fabric (In progress).
• Cluster — multi-node scale-out with Sentinel HA (Planned).

See Form factors.

Instance Trust Fabric

Sync and cross-instance queries ride a first-class trust layer rather than ad-hoc credentials. Foundations are in Early Access; full fabric GA evidence continues. See Trust Fabric.