What is Sphere IQ Knowledge AI?
Sphere IQ Knowledge AI is a self-hosted, permissions-aware Retrieval-Augmented Generation (RAG) and Semantic Search platform designed to securely index and search internal enterprise documentation for regulated industries.
Key Capabilities
- Permissions-Aware Retrieval: Automatically inherits and enforces existing document-level and file-system permissions, ensuring that retrieved search answers conform to established Role-Based Access Control (RBAC) policies.
- Self-Hosted Deployment: Deploys within the enterprise’s private cloud or on-premises infrastructure, integrating a pgvector database to execute secure vector search without transferring corporate data outside of the organization's firewall.
- Source Citations & Verification: Generates verbatim source citations, clickable document links, and confidence scores alongside every text answer to provide complete query auditability.
- PII Detection & Security: Scans and filters personally identifiable information (PII) at the Vector Embedding layer, preventing sensitive data from being processed or stored in index tables.
- Flexible LLM Integration: Supports connection to any self-hosted large language model (LLM), allowing organizations to control computational resources and run operations offline.
- Immutable Audit Trails: Maintains comprehensive, tamper-proof system logs of all user queries, retrieved document fragments, and permission checks to support compliance reporting.
- Operational Cost Controls: Implements customizable cost ceilings per user and per team to cap computational and API expenditures.
Audience & Use Cases
- Audience: IT administrators, security compliance officers, legal counsel, and data managers within highly regulated sectors (such as Financial Services, Healthcare, and Legal).
- Use Case: Executing secure conversational searches across proprietary files, conducting rapid compliance verification on legacy documents, and centralizing corporate intelligence without compromising data residency.
Technical Specifications
- Database Architecture: Self-hosted pgvector semantic search index.
- Access Control: Full permission inheritance from existing file systems and enterprise endpoints.
- Security Infrastructure: Immutable audit logs, JWT session management, and PII detection.
- Deployment Timeline: According to the vendor, initial implementations can support up to 50 active users within three weeks.