The exact AI stack I use for marketing in events

Most of what I write here is the operations side. Run of show, rooming lists, vendor trackers, the inbox. That is where a lot of the pain is, and it is most of what lands in your inbox from me.

But most of my job is marketing the events, not running them. Filling the room, getting found, and proving afterward that any of it worked. That half I talk about a lot less here.

So this is that half. The exact stack I use for marketing in events, what each piece does, and what it costs.

Nobody is paying me to name any of these.

Start here, and turn one thing off

If you’re using Claude, go to Settings, then Privacy, and turn off "Help improve our AI models." Leaving it on allows your conversations to be used for training. Before you point this at an attendee list or a sponsor contract, turn it off. Ten seconds, and it is on by default.

The base, and the number nobody mentions

Claude starts at $20, and Claude Code is included on every paid tier, including that one. The tiers differ in how much you can do per day, not in what you are allowed to do. I have watched marketers assume the useful version costs $200 and put the whole thing off for a year.

Claude Code is the part most people have never opened. The chat window answers you. Code acts on your machine and on your accounts. You will never read a line of code.

Getting found

Semrush has a connector, so I ask which terms bring people in and what competitors rank for without opening it. Search Console tells me what we already rank for and where we are slipping.

The more useful move is for tools with no connector at all. I use a service that shows what people are asking AI assistants rather than Google, and it has no way to get data out. So I taught Claude to drive the interface: it opens the site, reads the numbers, and brings them back. That works on any platform holding your own data behind a login and a dashboard, which in event tech is most of them.

Making the creative

Canva has a connector, so a deck or a graphic starts from an instruction rather than a blank page.

Higgsfield is where the image and video generation happens, and the reason it matters is that it sits in the same conversation where I write the promotion plan. The asset and the plan get made together instead of one waiting on the other.

Distribution

Beehiiv runs this newsletter, and it is connected, so Claude reads the archive when I draft a new issue. That is why the format stays consistent instead of drifting every time I write one.

For LinkedIn I use a scheduling tool with an API, which matters less than the habit it enables: the post and the follow-up that catches the people who engage get set up in the same sitting, rather than the second one never happening.

Website building easier than Squarespace

GitHub and Vercel, both free. This is why I can put a page live during a call. An event microsite, a one-page agenda, a speaker page that did not exist when the call started. Work that used to be a ticket sitting behind three other tickets I now do myself, while the call is still going.

Neon is the database underneath them, also free, and it is the piece marketers actually need.

You can tell anyone how many registered. Telling them how many walked in, which channel produced them, and what each one cost means joining three systems that have never spoken: ad spend, the registration source field, and badge scans. A spreadsheet is a file you open. A database is something you ask. The sponsor question, "how many of my target accounts were in the room," becomes something you can answer on the call.

One thing worth saying before you go and build any of this

AI is very good at scaling content. That is the problem.

If what you are making is generic, all of this will make you make more of it faster, and the only thing that changes is the volume. Volume was never the constraint. Nobody in your audience is short of things to read.

An agency owner in Calgary told me she spends more time editing AI copy than she would have spent writing it herself. That is what scaling content the wrong way feels like from the inside.

What makes a piece worth reading is that it knows something. A real number, or something that actually happened at your event.

Which is why every tool above is a connector rather than a chatbot. Each one is another source of things that are true about your business, and the more of them Claude can read, the less generic what comes out of it gets.

The stack is not there to help you write more. It is there so that what you write has substance.

The whole thing, in detail

Every tool with what it does and how it is wired in, the twenty-minute setup step that unlocks editing your own Docs and Sheets, the prompts I typed word for word, and what to do when it breaks in the first hour.

The stack →

Till next time,

Noah

Do More With Less Using AI

PS: If you’d like help setting this stack up for your business, reply to this email to learn more about how I could help you scale quality content using AI.