What it does

HASH is a multi-tenant platform that ingests raw data from public and private sources, structures it into a typed knowledge graph, and exposes it through visual interfaces meant for non-technical users. The core graph engine is written in Rust; the surrounding application layer uses TypeScript and Node. It supports bitemporal queries—tracking both when events happened and when the system learned about them—which is genuinely unusual in open-source databases.

The interesting bit

The “self-building” pitch: autonomous agents crawl sources, validate entities against schemas, and supposedly maintain the graph without constant human intervention. Whether this works at scale is unclear from the README, but the ambition is notable. The bitemporal support is the concrete differentiator—most graph databases treat provenance as an afterthought.

Key highlights

• Bitemporal data model with first-class support for event time and knowledge time
• Rust-based graph layer with TypeScript API and frontend
• Autonomous agent framework for data ingestion and validation
• Visual type and entity management for non-technical users
• Self-hostable via Docker Compose; managed cloud version exists but has a waitlist
• Browser plugin for capturing web data into the graph

Caveats

• Local development requires substantial resources: 8 GB RAM for Docker, ~15 GB disk, and 10–20 minutes for a cold Rust build
• The “self-building” claims are aspirational; the README offers no benchmarks or operational metrics to validate them
• Cloud signup currently involves a waitlist with manual review

Verdict

Worth exploring if you’re building systems where data lineage and schema evolution matter—supply chain, compliance, complex simulations. Skip it if you need a lightweight graph database or want something production-ready today without operational heavy lifting.