Unitag for AI assistants
You can give the assistant you already use a direct line to Unitag. Once it is connected, you can ask it a question about the product and get an answer from this documentation, ask it what is in your account, and ask it to do the work — create a QR code, change where one points, group a batch into a campaign.
There are two things to copy: the address of the Unitag connector, and the Unitag client ID your assistant asks for when you finalise the connection. Everything else is the sign-in you already have.
https://mcp.unitag.io/mcp
42deo5aspdubqjubemlal7iv54
The client ID is the same for every Unitag customer and is not a secret: it identifies the connector, not you. Access to your account still comes only from your own sign-in.
How the connection works
Unitag runs a connector built on the Model Context Protocol, the standard ChatGPT, Claude and Gemini all use to reach an outside service. You add its address once in your assistant's settings, and the assistant gains a set of Unitag capabilities it can use when you ask it to.
Access is granted by signing in, not by a key. When you first use the connector, your assistant opens the Unitag sign-in page; you log in with your normal account and approve the request. Nothing secret is ever pasted into a chat window, nothing sits in a settings file, and you can withdraw the access later without changing anything else about your account.
That is the difference between this route and the Unitag API. The API is for software you build; the connector is for the assistant you already have open.
Choose your assistant
The connector is the same in all three. What differs is where the setting lives, and — for Gemini — which of Google's products can hold a third-party connector today.
What you can ask for
A connected assistant can do three kinds of thing.
Read the documentation. Every page of this site, in English and in French, as source material rather than a search result. Answers cite the page they came from.
Read your account. Your QR codes, campaigns, domains and vCards, and the scan data behind them — how a campaign performed last month, which codes have not been scanned at all, which domain is still waiting on a DNS record.
Create and change things. Make a QR code, edit a destination, rename or re-group a batch. These run as you, so they take effect in your account straight away.
Questions that work well
The assistant is best at the jobs that are tedious by hand — the ones that need one answer drawn from many screens.
"How do I put a frame with a call to action under a dynamic QR code?"
"List the QR codes in my Autumn campaign with their scan counts for last month, worst first."
"Create a dynamic QR code pointing at our new product page, name it Autumn flyer, and add it to the Autumn campaign."
"Three codes on the old catalogue URL are still live. Point them at the new one."
"Which of our branded domains is not fully verified yet, and what is missing?"
What the assistant can and cannot do
The connector does not create a second, wider way into your account. It gives your assistant the account you already have.
It acts as you. Same organisation, same role, same plan. If your account cannot see a sub-organisation or edit a domain, neither can the assistant.
It only acts when asked. Reading is routine; anything that writes is a step the assistant asks you to approve first. Read what it proposes before you accept it.
There is no key to leak. The client ID is public and grants nothing on its own. Access comes from a sign-in you can revoke — from your assistant's connector settings, and from Unitag's side — without rotating anything or breaking another integration.
Everything it does is on the record. Actions taken through the connector appear in your activity logs like any other, with the account that made them.