Features that put you in control.

Every capability below is described by what it lets your organization do, not by marketing adjectives.

Start trial Book a demo

Stylized ocatis governance console: conversation sidebar, chat panel with model selector and usage chart, and a policy strip with Allow, Ask and Deny toggles

Multi-provider access

Native provider support for OpenAI, Anthropic, Google Gemini, Azure OpenAI, Amazon Bedrock and Google Vertex AI
Any OpenAI-compatible endpoint, plus local inference via Ollama or LM Studio
Personal providers so users can add their own endpoints within admin-set boundaries
Model catalog synchronization with per-model settings: display name, context window, vision, tools and reasoning toggles

Model governance

Model enablement rules per tenant, group and user
Reasoning and thinking rules per system model
Tenant-wide defaults for systemic tasks such as title generation
Personality system with configurable sampling parameters

Credential security

AES-GCM encrypted storage for API keys and MCP secrets
Tenant-scoped or per-user key assignment with role-based administration
Real revocation instead of key rotation by copy-and-paste
Access and refresh token authentication with RS256-signed JWTs

MCP tool governance

MCP server management with encrypted connection secrets
Per-tool policies: Allow, Ask, Deny
Approval mechanics for tool execution, conversation-scoped tool configuration
Managed MCP servers at system or tenant scope, sandboxed in Docker containers

Conversation workspace

Streaming chat with reasoning visualization and tool-call display
Conversation branching, message deletion with preview, chat search
Folders, tags and pinned conversations for fast access
Per-conversation token statistics with a context-limit indicator
File upload and download in conversations

Users and tenants

Global user identity with access to multiple tenant workspaces and fast workspace switching
Users, roles and groups with role-based administration
Email and password plus Google OAuth sign-in, email verification on signup
Profile images with upload and Gravatar fallback

Languages and layouts

Multi-language client including RTL support
Adaptive layouts for mobile, tablet and desktop
Cross-platform client: web, iOS, Android, macOS, Windows, Linux

Usage and analytics

Platform usage report for system administrators
Tenant-level aggregation in the usage analytics dashboard
Per-conversation token statistics for cost and context control

Organizational memory

Retivum knowledge graph: entity-to-entity facts with bi-temporal validity — what was true when stays queryable
Provenance on every fact back to the source episodes that surfaced it
Hybrid retrieval merging full-text, vector and graph traversal with point-in-time as-of queries
Decision records with reasoning, outcomes and causal or precedent links inside retrieval results

Read the full organizational memory story →

Next: Architecture →