Agent layer for SaaS with grounded answers, sandboxed tools, and per-customer isolation
High-assurance infrastructure for AI agents
Stage
Not given
Not given
Founded
Not given
Not given
Based in
Not given
US
Pricing model
Free
Open source
Pricing
Free during launch period. Customers pay their LLM provider directly for model usage at provider rates; Oberik does not markup or resell tokens. No credit card required, no usage meters tracked by Oberik itself.
Not given
Key features
Document-grounded answers with page and quote citations
Linux sandbox environment for agent execution
Per-customer cost tracking and billing
Memory and wiki persistence across sessions
Scheduled and recurring work with durability
Capability-scoped JWT tokens enforced per-request
Support for custom tools and MCP servers
Multi-model support with bring-your-own-keys
Server-side prompt injection and PII screening
SSH dashboard interface for programmatic configuration
Harpe: chooses what an AI agent reads under a token budget with consistent results
Kibisis: keeps agent memory on your hardware with AES-256-GCM encryption
Argus: append-only record that detects edits with cryptographic verification
No network code by default for local and disconnected use
MIT License with published code, benchmarks and raw results
What makes it different
Hard isolation between customers with no shared state or tenant leakage
Citations with document, page and quote provided automatically with every answer
Only two API calls needed to integrate grounded AI into any SaaS product
Customers never charged by Oberik; direct billing from their own LLM provider keys
Measured empirical validation across 6,111 questions outperforming full-context models
Zero plaintext on disk with encrypted vaults matching plaintext accuracy
78.3% fewer tokens sent to the model with a 1,200-token budget per turn
Preserves conversational chronology and relational qualifiers under token budgets
Oberik
Perseus Computing