XONTAINERLogistics ecosystem
TREN
Early access

Connect your AI to your operation

Your AI runs your operation, your team keeps the last word. You connect the AI you already use, through MCP or a command-line tool, and it works only inside the role and the company you give it.

Status: early access This is not live yet. The connection layer is being built in stages, and the sequence is below. If you want to be among the first companies to try it, use the early-access form and write “AI connection” in the notes.

Early-access form

How it works

  1. 01

    Register

    Your company is registered and verified. Without a verified company there is no connection.

  2. 02

    Connect

    Your AI client asks to sign in at id.xontainer.com using OAuth, and the user approves. The MCP server will be at mcp.xontainer.com (planned).

  3. 03

    Assign a role

    Each user has a role. The AI sees only the tools that this role and this company allow, and nothing else.

  4. 04

    Work

    The AI reads, suggests and opens requests. Every call checks company and role again and is written to the audit log.

What sits between your AI and your data

Your AI cliente.g. Claude, ChatGPT, Gemini, Cursor id.xontainer.comOAuth sign-in, user approves mcp.xontainer.comMCP server (planned) Company + role checkon every call, logged TMS toolsWMS toolsCustoms desk tools Approval cardA person approves in the app Your AI cliente.g. Claude, ChatGPT, Gemini, Cursor id.xontainer.comOAuth sign-in, user approves mcp.xontainer.comMCP server (planned) Company + role checkon every call, logged TMS toolsWMS toolsCustoms desk tools Approval cardA person approves in the app
Illustrative flow. The server addresses are planned and not live yet.

Three levels of what the AI may do

Not every action carries the same risk, so the AI does not get the same freedom everywhere.

Three levels of what the AI may do
LevelWhat the AI doesExamples (illustrative)Who has the last word
ReadAnswers questions from your data, within the user’s role. Free to use.
  • Example Which of my trucks are waiting at the border?
  • Example Who is arriving at the dock tomorrow?
  • Example Which T1 guarantees are close to expiry?
Nothing is changed, so nothing needs approval.
SuggestPrepares a draft. A person approves it with one tap in the app.
  • Example Draft a declaration from this invoice.
  • Example Open a trip draft for this load.
  • Example Suggest a dock appointment.
The person who approves the card.
CriticalPayment, invoicing, release from bonded storage, prices and customer master data. The AI can only open a request.
  • Example Request release of this consignment.
  • Example Request a payment run.
Two human approval steps. Automation only with measured accuracy and your company’s approval.

Three ways to bring the AI

Where the model comes from is your choice. The security rules are the same in all three.

aXontainer AI
Model access that we provide, bought as a credit package. When the spending limit you set is reached, it stops.
bBring your own key
Your company uses its own account with a model provider and connects it.
cOpen models on your own server
Open models run on your premises or on a server reserved for your company alone.

The security model

These rules do not depend on which AI you connect.

Verified company first
No registration and verification, no connection.
OAuth sign-in with consent
The AI asks, the user approves at id.xontainer.com. The AI never receives a password.
Only the tools your role allows
The AI sees the tools that the user’s role and company permit, not the whole catalogue.
Check on every call
Each call checks company and role again. One company’s AI cannot see another company’s records.
Audit log
Every call is recorded: who, which tool, when.
One-tap disconnect
Your company can cut the connection at any moment.

Order and status

Honest status: all of this is planned, in this order.

  1. Phase 1

    Read tools

    Questions answered from your data, first for TMS, then for WMS.

  2. Phase 2

    Suggestions and approval cards

    The AI prepares drafts; a person approves in the app.

  3. Phase 3

    Customs desk tools, CLI and automation permission

    Tools for the customs desk, the xontainer command-line tool, and a per-company permission for automation.

New to MCP? Read the short guide: How to connect your AI to your logistics operation (MCP)

Questions about connecting your AI

What is MCP?
MCP (Model Context Protocol) is an open-source standard for connecting AI applications to external systems. Through it, an AI client can use data, tools and workflows that a server offers.
Can I connect my AI today?
Not yet. This is early access: the connection layer is planned and will arrive in stages, starting with read tools. You can ask to join early access through the form.
Which AI clients will work?
Clients that support MCP, for example Claude, ChatGPT, Gemini or Cursor. These are examples; no partnership with these products is implied.
Can the AI change things without a person?
Not by default. Reading is free within the role. Suggestions wait for one tap of approval. Payment, invoicing, release from bonded storage, prices and customer master data need two human approval steps, and automation needs measured accuracy and your company’s approval.
Does the AI see all my company’s data?
No. It sees only the tools allowed by the user’s role and company, and each call is checked again and logged.
Who pays for the AI?
It depends on the way you choose: Xontainer AI with a credit package and a spending limit, your own provider key, or open models on your own server. Prices are not published on this page.

Want to be among the first to connect?

Leave a short note in the early-access form with your role and operation, and write “AI connection” in the notes. The form is a prototype for now and sends nothing.

Early-access form

Start from your role:Customs broker Carrier Warehouse / 3PL Shipper