ZenCTRM™

AI agents with their own identity and permissions.

ZenCTRM's trading, valuation, settlement and configuration functions are agent tools through the Model Context Protocol (MCP). An AI assistant, for example Claude Desktop, works with only the permissions of the person it acts for.

What an agent can do through MCP.

The main ZenCTRM functions are MCP tools. Ask in plain words and the assistant chooses the tools. Each tool needs the same permission as the screen it mirrors.

  1. Book and manage trades

    Book from a desk template, amend, move a trade through its lifecycle and record deliveries.

    "Book Buy Lagoon CPKO for 2,300 MT, April 2027, at 5,460 MYR."

  2. Price, value and explain

    Read curves, preview calculated curves, value the position, store an end-of-day run, check that it reproduces and explain the P&L.

    "Value the palm oil book as at yesterday."

  3. Settle, invoice and confirm

    Generate cash flows, run settlement, draft invoices and credit notes, produce confirmations and export the journal.

    "Draft September's invoices for Sumatra Agri."

  4. Configure, as drafts

    Add a field, define a custom object or write a validation rule. The agent checks that it works. A person switches it on.

    "Add a vessel name field to physical trades."

Two identities on every action.

Every change an AI assistant makes in ZenCTRM is a new version of the trade, recorded as done for a named person. The agent can take only the steps its person may take, so a step that needs a second person still needs one.

  • Acting for a named person

    The agent holds only what both it and the person hold.

  • Refusals in words

    "This needs the Invoice Issue permission." The agent tells the person why.

  • Your vocabulary

    Agents read a catalogue of your business operations, kept by your people.

Watch an agent book a trade under your controls.

A 30-minute session. Bring a real trade and your approval rules.

Request a demo