Skip to content

INDB — Interpretational Database

INDB = Interpretational Database.

A new class of system — alongside relational, document, graph, and vector stores — defined by one rule:

Meaning is not stored at write time. It is produced at read time.

Not a database. Not an archive.

Classical DB / archive Interpretational DB (INDB)
Unit Row / document Event (anchor, not conclusion)
Write Record final state Inhale raw observation
Read Fetch what was stored Interpret, Prism, Echo, Subwave
Growth More data = heavier Fusion + forgetting (salience, cold tier)
Contradiction Error or overwrite Signal (perception_gap, paradox)
Privacy Access control Black Box + signed presence

Archive preserves the past. Memory compresses, fades, and wakes on resonance.

Interpretation stack (API surface)

Phase Endpoints / engines
Inhale POST /events, UDP, gRPC, MCP ingest
Exhale Fusion, TTL, entropy prune → sleep (not erase by default)
Resonate Echo, Subwave, horizontal graph + index
Interpret POST /interpret, Prism, Deduction
What-if POST /slice, POST /what-if (core = no LLM)
Axiom Ed25519 on responses; blind_payload contract

LLM (Hermes/Ollama) is optional voice — not the interpretational core.

One sentence

INDB is an interpretational database: signed cognitive memory where context, observer, and «what if» reshape meaning without rewriting facts.