The msg2ai Toolkit: Set Up a Conference Assistant by Asking

Why Conference Setup Eats the Week Before the Event

Event teams routinely move the same information between systems by hand: registrations out of a ticketing platform, into a spreadsheet, into a messaging tool, sometimes into a badge printer on top of that.

• Registrations get exported and re-imported — manually, often more than once as numbers change in the final week
• Agenda and speaker details get entered twice — once for the event website, again for the messaging or app tool
• Check-in and scheduling live in separate tools — so the team can't see arrivals and meeting requests in one place
• Feedback goes to a fourth tool entirely — disconnected from the registration and messaging data that would give it context

The busiest week of an event's life is the week before it, and most of that time goes to moving information between systems rather than preparing the event itself. The msg2ai Toolkit moves that work into a conversation instead.

What the msg2ai Toolkit Actually Is

The Toolkit is a small connector you install once. It gives an AI assistant a safe, limited door into your msg2ai account.

It comes in two forms, and most event teams only need the first:

• A connector for AI tools such as Claude, so you can simply describe what you want
• A command line version, for teams who prefer to script a repeatable event setup

There are four versions, one for each kind of business: hotels, events, trips, and vacation rentals. Each one speaks the language of that business. Behind all of them sits the same set of seventy capabilities, covering assistants, knowledge, contacts, campaigns, surveys, meetings, check-in, and reminders — for an event, that means the agenda, ticketing sync, check-in codes, meeting scheduling, and post-event surveys in particular.

Not a Generic Chatbot Builder

The assistant isn't guessing or writing filler copy into a void. It works with the real records already in your account. That means it can:

• Read your existing assistants and attendee lists
• Pull knowledge from your own event website
• Create and update real records you can see in the msg2ai app
• Report back exactly what it did

Everything it creates appears in msg2ai exactly as if a person had entered it. You can review, edit, or undo any of it.

Example: Running a Six-Hundred-Person Conference

An event team is preparing for a six hundred person conference and is currently moving attendee lists between several systems by hand.

After the doors close, a feedback survey can be created, opened, and later closed, with results summarized for the debrief. None of these steps is hard on its own — the point is that they normally live in six different places, and the Toolkit lets one assistant assemble them from the systems that already hold the answers. It's the same messaging-first approach behind how SAM boosted conference engagement with just a text.

A Second Example: Reaching a Segment Before Doors Open

This is the simplest example, and often the one teams reach for first once the event assistant is live. You can ask it to show who's registered today, build a segment — VIP pass holders, or attendees who haven't checked in yet on the morning of day two — write the message, assemble the announcement, and show it to you before anything is sent.

Only then does sending happen, and only if you ask for it. Campaigns can also be paused or cancelled after they start — useful when a room change or delay needs to reach only the people it affects.

You Decide What It's Allowed to Touch

This is the part worth reading twice, because it's what makes the rest comfortable. Access is granted through a key that an administrator creates in Settings. When you create it, you choose exactly what it can do.

• Keys are read-only by default
• Writing, sending, and exporting are each granted separately
• Keys expire, and can be revoked at any time
• Every key has a daily limit on how many attendees it can message
• Phone numbers in contact and conversation records are shown only as the last four digits

In practice this means you can give a drafting assistant a key that can prepare an entire announcement and is physically unable to send it. Sending stays with a person, or with a key you deliberately trusted. Because sending costs money, it's treated carefully: recipients are counted against the key's daily allowance before anything is queued, and a send that goes over the limit is refused rather than partly delivered.

What Still Needs a Person

A few things are deliberately left out, and it's fair to say so plainly.

• Getting a brand new phone number is still handled by our team. The Toolkit can move a number you already have between your own assistants, but it can't claim a new one
• Submitting message templates for WhatsApp approval still follows the normal process
• Changing an assistant's core instructions requires a key an administrator has specifically allowed to do that

These aren't gaps waiting to be filled quietly. They're the decisions we think should stay with a named person.

Getting Started

If you already use msg2ai, you can try this today.

That last step is the honest way to begin. Before letting anything create or send, ask it what it can see. It's the fastest way to build trust in what you've just connected.