AI agents

AI agents that work under your rules.

Agents act through the same rules as your staff, never around them, and you approve what matters.

What the ledger enforces

  • One posting engine for every journal
  • Fiscal period locks
  • Hash-chained, tamper-evident audit trail
  • Strict per-company data isolation

How agents are kept in check

  • Schema-validated agent tools
  • Dry-run simulation before anything is posted
  • Human approval above set thresholds
  • Scoped credentials and rate limits
  • An event feed of every action taken

Connecting an assistant

  • Link your own assistant, such as Claude or ChatGPT, through our MCP server
  • It signs in with your account and can do no more than you can
  • Use your own AI provider's key

Built-in assistants

  • Ask about the business in WhatsApp or Slack, in Arabic or English
  • Approve requests from the chat with a button
  • Multi-step playbooks with one preview and one approval
  • Supplier bills read from a PDF or e-invoice file and matched to the order

See Univia on your own processes