Skip to content

classify → route → block → record

Sensitive data never leaves your control. Automatically.

Every message is classified before it leaves and sent only to providers approved for its level — frontier models for everyday work, your own models for restricted content.

Incoming message

"Can you check these figures against client account GB29 NWBK 6016… before Friday?"

Classifier

Restricted

method: pattern
reason: bank account number

PUBLIC · INTERNAL

Frontier models

Claude · OpenAI · Gemini · Groq

not allowed for this level

CONFIDENTIAL

Providers you approve

e.g. Claude under your agreement

not allowed for this level

RESTRICTED

Your own server

Ollama · runs on your infrastructure

✓ routed here

No approved provider online? The message is blocked — never downgraded.

Every decision, on the record

Level, reason and provider are stored with each message. Users see why; admins inspect and tune patterns, examples and thresholds for their own organisation.

Evidence for clients, auditors and your DPO.

{
  "classification": "restricted",
  "method": "pattern",
  "reason": "bank account number",
  "routed_provider": "ollama",
  "auto_selected": true
}

Who it's for

Teams trusted with other people's data

Your team already uses AI. Lexix makes the safe choice the easy one — so client data stops ending up in personal accounts.

Accounting & advisory

Client financials stay with providers you've approved.

Law & legal teams

Privileged material never reaches a model you haven't cleared.

Agencies

Work under client NDAs without banning AI.

Fintech & health-adjacent

PII and credentials caught before they leave.

Built for the rest of the day, too

Multi-model chat

Claude, OpenAI, Gemini, Groq and Ollama with streaming responses.

Company-wide context

Set it once; every conversation starts with it.

Governed MCP tools

Org-wide servers with per-user tool permissions.

Bring your own keys

Encrypted at rest, masked in the UI, scrubbed from logs.

Attachment PII checks

Files are scanned before they reach a model.

Usage & budgets

Token spend per provider, with budgets.

All features →

Give your team AI you don't have to worry about.

Free during beta. Bring your own API keys.

Start free