Features
Control first. Then everything else.
Every message is checked and routed by sensitivity before it leaves — wrapped in a workspace your team will prefer to their personal AI accounts.
CORE
Automatic sensitivity routing
Each message is classified as public, internal, confidential or restricted and sent only to providers approved for that level. No approved provider means blocked, not downgraded.
CORE
A record of every decision
Classification, reason and provider are kept with each message. Users see why; admins can inspect it.
CORE
Your own rules
Add patterns and examples — client names, project code names — and tune thresholds. They apply only to your organisation, on top of the defaults.
Multi-provider chat
Claude, OpenAI, Gemini and Groq for everyday work, your own Ollama models for restricted content. Streaming responses.
Company-wide context
Teach Lexix how your company works once. Every conversation starts with it.
Governed tools
Connect internal tools and services via MCP, and choose per user which tools they can use.
Attachment PII checks
Uploaded files are scanned for personal data before they reach a model.
Conversation history & folders
Keep work organised and searchable with persistent history and folders.
Usage & budgets
Track token usage and cost per provider, and set budgets.
Feature overview
| Feature | Beta |
|---|---|
| Data Protection | |
| Automatic sensitivity classification | ✓ |
| Routing by sensitivity level | ✓ |
| Block when no approved provider | ✓ |
| Attachment PII detection | ✓ |
| Routing decision per message | ✓ |
| Team & Users | |
| Team members | Unlimited |
| User roles & permissions | ✓ |
| Team invitations | ✓ |
| AI Providers | |
| Claude (Anthropic) | ✓ |
| OpenAI | ✓ |
| Groq | ✓ |
| Ollama (Local) | ✓ |
| Gemini | ✓ |
| Features | |
| Company-wide context | ✓ |
| Conversation history | ✓ |
| Conversation folders | ✓ |
| MCP tools with per-user permissions | ✓ |
| Usage tracking & budgets | ✓ |
| Support | |
| Community support | ✓ |