Hi,

Ask a room of event professionals whether they use AI and almost every hand goes up.

Ask the same room to explain the difference between a prompt, a workflow and an agent, and the hands come down.

That gap is not a knowledge problem. It is the reason most AI work in events stalls at the same point: useful, but never quite load bearing.

The step everyone skips

There is a pattern to how this usually goes. Someone decides a task should be automated. They open a tool. They start building.

And it half works, in the way these things half work. It produces something. It saves a little time. But it never becomes the thing you trust to run without you, and eventually it quietly stops being used.

The step that got skipped was the one before the tool. Nobody mapped the workflow.

An agent is not a clever prompt. It has a structure: a trigger that starts it, the inputs it needs, the knowledge it can draw on, the steps it takes, and the output it has to produce. If you cannot describe those five things about a task, you cannot automate it, and no tool will rescue you from that.

Most of the difficulty in building an AI agent is not technical. It is that the process being automated was never actually written down.

What Thursday actually is

Episode 3 of the Events AI Playbook is a working session, not a talk.

Paula Rowntree takes one real event workflow, comparing venue RFPs, and maps it in front of you into the structure an agent is built on. You will learn how to tell a prompt, a workflow and an agent apart in plain English. You will learn how to judge whether a task is worth automating at all, which is the question most people never ask. And you will leave with a mapped workflow rather than a set of notes.

This is Part 1 of a two part build. In Episode 4 on 19 August, you take the workflow you mapped on Thursday and build the agent from it. Which means Thursday is not optional if you want August to be useful.

Thursday 23 July. 11:30 a.m. SGT. 1:30 p.m. AEST. Live, with Paula.

If you are joining late: two episodes are already in the archive. Episode 1 maps the AI tool landscape for event work. Episode 2, Prompting For Power, is the one people tell us they went back to — if you only have time for one before Thursday, watch that one.

You have already seen this structure somewhere

Here is the quiet connection worth noticing. A trigger. Inputs. Steps. An output. That is how an agent works - and it is also how the Summer Release, arriving later this week, has been built.

An event ends, and it archives itself out of your active workspace. An attendee checks in, and the pages they can access adjust to match. A scheduled onsite meeting passes its time slot, and participants are asked whether it actually took place. Across Registration and Omnichannel, the release teaches the platform to watch for a moment and act on it, without anyone having to remember to.

There is more in it than we can preview here - a redesigned event management workspace, configurable social sharing for event links, a faster and more secure hosted checkout, and the launch of IcebergHub, an AI-powered advocacy platform built with The Iceberg Charity Trust. The full details land on our Product Updates page this week, and we will walk through them properly in an upcoming issue.

But the pattern is the point. The thinking you will practise on Thursday is the same thinking arriving in the platform you already use. Map a workflow on Thursday, and you will recognise the structure the moment the release lands.

See you on Thursday.

We will be in your inbox every Tuesday.

The Gevme Team