Research program
LoomDB
A graph-based temporal memory architecture.
Active research · Proprietary
agent memory · retrieval · temporal graphs
Rust · WASM
The question
Can memory retrieval be modeled as spreading activation over a temporal graph — where forgetting is a feature, not a failure?
Overview
Context as a living graph: activation spreads, attention decays, relevance is a function of time. Memory that behaves less like a database and more like a mind — recency, frequency, and association shaping what surfaces and what fades.
Approach
- 1.Activation decay: memories lose energy over time unless reinforced, mirroring human salience.
- 2.Active context graph: the working set is a live subgraph, continuously re-weighted as the conversation moves.
- 3.Rust core compiled to WASM — the same memory engine runs server-side, in the browser, and embedded in desktop agents.
Open problems
Stated plainly, because pretending they're solved would be the opposite of research.
- Consolidation: when should episodic traces merge into semantic structure?
- Decay calibration across radically different interaction frequencies.
- Benchmarking long-horizon recall without long-horizon datasets.