SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
Hillock is a 100% local neurosymbolic memory engine built for edge hardware. Its 3-layer architecture combines a relational knowledge graph, Hebbian synaptic memory, and vector symbolic structures. Document parsing and similarity gating run on the device without cloud dependencies or API costs; a local Ollama LLM is used only to generate the final response.
KEY FEATURES
Key features
- 3-layer neurosymbolic memory architecture
- On-device document parsing and similarity gating
- Combination of a relational knowledge graph and Hebbian memory
- Optional use of a local Ollama LLM only for final responses
EDITOR'S NOTE
Why did it stand out?
It separates document ingestion and query decisions from the generative model and handles them through 3 local memory layers. A strict gate passes facts to the final-response LLM only when they exceed a threshold.