On-prem AI memory with anonymisation, tenant isolation, PQC-ready signing and full audit.
No anonymiser, no cloud call. No log, no serve.
Post-quantum signatures; per-tenant seal isolation.
Sealed release & exchange provenance, verifiable.
Local Ollama or your keys; MCP; self-hosted control.
The technical capabilities behind this — from the live feature inventory.
Each install continuously monitors and defends its own host across Linux/Windows/macOS guests and internet servers. Event-driven detection triggers reversible containment within seconds — ban a brute-force IP, close a rogue public port, isolate a tunnel, patch within SLA — while irreversible actions stay behind an operator gate. Bounded by scoped, expiring, revocable policy + a circuit breaker + a control-plane invariant that never locks SSH + a global kill-switch. 108 tests, shadow-verified on production; autonomous actions ship in phased enablement.
Three-tier support built into the product: local self-repair (L0) → RAG-over-docs hub expert (L1) → signed, operator-approved remote intervention (L2). The MSP support model ships inside the product.
Data is anonymised at write and again at the LLM proxy by provider trust-level, so the cloud model never sees real PII; de-anonymisation is on-demand and fully audited.
ML-KEM-768 (NIST FIPS 203) at rest, per-project Fernet rotation, threshold key escrow; per-tenant seal isolation; RBAC enforced at query level.
Qualified-sealed release & exchange provenance (RFC-064) — every published bundle carries a verifiable ed25519 signature and a qualified seal.
Free engine on your own hardware. No credit card. Founder pricing through 31 October 2026.
View pricing Back to home