Who Actually Turns Up to AI Meetups in Singapore

I run a free AI meetup in Singapore called AI Power Users. Seven events in twelve weeks, between May and July 2026. Every one of them filled.

Somewhere around the fourth event, I got curious about something I had only been guessing at. Who is actually in this room?

I had an assumption, and I suspect you have the same one. An AI meetup sounds like a room full of developers.

It is not. Not even close.

Only 6.2% of the room works in engineering

I pulled the registration exports for all seven events and de-duplicated them. That gives 625 unique people and 953 signups across 495 different organisations. 580 of those people gave a job title.

Here is how the titles break down.

Role	Share
Founder, owner or CEO	18.6%
Manager or director, discipline not stated	16.2%
Marketing and comms	7.5%
Data, analysis and research	6.4%
Engineering and technical	6.2%
Sales, BD and agents	6.1%
Product	5.3%
Consulting and advisory	4.5%
HR, L&D and education	4.3%

Engineering sits fifth. It is behind marketing, behind data, and roughly level with sales. Founders outnumber engineers three to one.

That single number has changed how I plan every event. When you believe you are teaching engineers, you reach for the technical thing. When you know that 94% of the room does something else for a living, you reach for the workflow instead.

It is a senior room, and that surprised me more

Seniority splits like this:

• Founder or C-level: 22.4%
• VP, head or director: 15.2%
• Manager or lead: 15.8%
• Individual contributor or practitioner: 20.6%

About 38% are director level or above. People from GovTech, DBS, Grab, NUS, IMDA, SMU, TikTok and FairPrice have all sat in the room, alongside a very long tail of small companies. 495 organisations across 625 people means most attendees are the only person from their company in the building.

That last detail matters more than it looks. These are not teams sent for training. They are individuals who decided on their own that they need to work this out, and who are senior enough that nobody is going to tell them how.

They already pay for the tools

At two of the events, I asked what people were already using. Between 85% and 88% were already paying for ChatGPT or Claude. Only about 6% said they planned to subscribe before the event.

This is the finding I would most want any other organiser to take away, because it invalidates a whole genre of content.

If you are running an AI session, the temptation is to start at the beginning. What is a large language model, how do you write a good prompt, here is how to sign up. For this audience that opening is dead air. They have been paying for the tool for months. What they do not have is a way of working with it that survives contact with a real job.

The gap is not knowledge. It is depth.

That is also why the people teaching are mostly not me. Members who have worked something out in their own job come up and show the room how they did it, which lands differently from a demo by someone selling the tool.

About 55% of people who get a seat actually turn up

Three of our events had check-in properly run. The show rate was 56.0%, 54.4% and 58.1%.

Three independent events agreeing within four points is stable enough to plan on. If you want 60 people in the room, approve about 110. Measured against everyone who signs up rather than everyone approved, it is closer to 43%.

I should be honest about the limits here. Check-in was not run properly on three of the other nights, so I genuinely do not know how many people were in those rooms. I would rather publish four solid numbers than seven soft ones.

One sub-group is worth calling out. 20 students registered across the series. Two turned up. That is by far the worst show rate of any group, and if you run a free event with limited seats, it is worth knowing before you approve.

A third of them come back

203 people, about a third, have signed up more than once. One person has signed up for all seven.

At the same time, 90 of the 136 signups at the most recent event were first-timers. Twelve weeks in, most of the room is still new. Both things are true at once, and the combination is what makes the events worth running: a returning core who set the tone, and a majority who have never been.

The number I am least happy about is this one. 140 people have never got into a single event. They raised their hand, the room was full, and they were turned away. That is more than a fifth of everyone who has ever tried to come, and no amount of marketing fixes it. Only a bigger room does.

What I got wrong

Three things I believed at the start that the data disagreed with.

I thought it was a builder community. I used the word “builders” in early copy. It is dev-coded, and it was quietly filtering out the operations, finance, marketing and HR people who turn out to be the majority of the room. I stopped using it.

I thought the audience was solo founders. Early on it was. At the most recent event, 41.5% came from companies of 50 people or more, against 22% who are “just me”. The room is drifting toward employed professionals inside larger organisations, which is a different set of problems. Getting a company to adopt something is harder than adopting it yourself.

I thought everyone came for the same reason. At the last event I asked, and the room split four ways: 35% want to break into a tech role, 24% already work in one and came to compare notes, 27% were curious how tech teams work, and 13% are hiring. That is not one audience. It is at least two, and eventually they will want different nights.

If you are running something similar

Four things I would do differently from the start.

Capture the fields you will want later. I have no industry field, so 59% of company names tell me nothing about sector. I have no gender field, so I cannot answer that question honestly at all. Both are one dropdown at registration.

Tag your links. The UTM source is blank on all 953 signups. I cannot tell you which LinkedIn post, which partner or which referral drives registrations. It is the cheapest fix available and I still have not done it.

Run check-in properly every time. Three events with no reliable count means three events I can say nothing about. That data is not recoverable later.

Ask one question at registration. We ask what people most want to get out of the night. It shapes the run of show, it tells you which topics are live, and after seven events it is the single most useful thing in the export.

The short version

The people showing up to AI meetups in Singapore are not the people you would picture. They are senior, they are commercial, they already pay for the tools, and they are mostly not engineers. They are marketing leads, founders, sales directors and operators, trying to work out what this means for a job they already have.

If that sounds like you, that is the room.

AI Power Users is free, it runs fortnightly in Singapore, and sign-ups go out in the WhatsApp group first. You can see every past meetup and join the group here.