How many AI superusers do you need? Research says one per team.

August 26, 2026

Hey Community!

AI adoption and sophistication across teams can be uneven, but it turns out that everyone doesn't need to be an expert. All it takes is one, according to a recent experiment.

Twice a year, Atlassian runs a hackathon called ShipIt, where employees form small teams and have 24 hours to build something.

My team, the Teamwork Lab, compared project judging scores against internal AI usage data for over 2,500 participants across 1,200+ teams. We found a big 0→1 effect: teams with one superuser were around 18% more likely to be top-scoring.

It just takes one superuser_0-1 effect.png

So even if AI adoption isn’t evenly distributed across your team, just one superuser can make a big difference. Who's the superuser on your team?

Read the full research!

2 comments

I recommend adding "per team" at the end of the second sentence of your headline. It is more precise to the then following content and it's something that I can say we as a company group are working towards.

Whenever we install AI and/or automations into a process we always make sure to elevate at least one person - and if reasonable a backup - to the superuser level and as someone that can not only be the first supporter, but a contact person that speaks both languages: process team and AI team.

Like • like this

Great call, @Richard Scholtes, on the headline - it should be more clear now!

I like your approach, having a first AI responder/supporter and when possible, a back-up. I hadn't thought about this superuser needing to speak to both the operations and tech but yes, that makes a ton of sense.