)
)
TL;DR
We’ve told you what the Factory Rebuild is, now here’s the how! Duncan (our Founder and CEO) and Sophie (our COO) get honest about what the rebuild actually took, including:
Why "do you use AI?" is a useless question,
The split that opened up between the makers and the unlearners,
Why being ‘wishy-washy’ with comms is expensive,
Why you can’t change the whole team all at once,
The fact you'll run two worlds at once,
Why you need to kill the "same thing but cheaper" myth ASAP.
The AI models were the easy bit.
When you picture a software agency going AI native, you probably picture the tech. New models, new tools, a room full of engineers wiring it all together.
We thought the hard part would be the tech too. We've got a building full of very technical humans who really know their stuff, so surely the AI models were going to be the tricky part?
Boy, were we wrong!
Turns out, the AI models were the foothills. The mountain being to ask fifty-odd people to unlearn their way of working, some of whom have worked that way for fifteen years. All while still shipping client work to deadline.
"Transforming an organisation involves transforming people as well as the technology. And transforming the people is the hardest thing."
Duncan Cook, Founder, 3 Sided Cube
Catch the full conversation
New here? Our Factory Rebuild series is us sharing our honest story of going fully AI native. This is the first stop on the journey, straight from the two people who lost the most sleep over it 💤
Learning 1: "Do you use AI?" is a big question
Early on we'd ask people whether they were using AI and everyone said yes. Great! But what we didn’t really realise is that asking “do you use AI” is a lot like “how long is a piece of string”. Usage stretches across a spectrum the size of a solar system (promise we’re not exaggerating).
One person drafts the odd email with it. Another has it wired into how they build and test. Same question, same “yes” answer. Yet, keeping with the solar system theme, these are on two entirely different planets!
"Would you build a world class running team by just asking if you run?"
Duncan Cook, Founder and CEO, 3 Sided Cube
What we’d do if we had a re-do
The question was never whether people touch AI. It's how far in it actually goes. Instead, we’d skip the yes/no entirely. Focus on asking what someone had actually built with it, and what they'd stopped doing. Then we'd share the real examples of what deep use looks like and let people place themselves against it.
)
Learning 2: You'll have makers and unlearners,
And they want opposite things 👀
One thing we didn't see coming was the non-developers taking off like rockets!
People who'd spent years working in tech, always one step back from the actual building, were suddenly the ones building. Our Project Managers, People team, Marketing team, and more, all went from asking developers to step in, to building their own tools. None of them developers. No one would’ve predicted this a year ago.
On the other side, the developers had the steeper road. Not only were they picking up new tools, they were also putting down years of instinct. Then there's the layer underneath. Identity. When writing code has been how you’ve added value for your whole career, and that part gets easier, something shifts that no new process can paper over. It stops being a question about your workflow and turns into a question about you.
Where's the worth now? It’s still very much there. It’s just taken up a notch into the architecture, the calls you make, the taste to tell good from great.
What we’d do if we had a re-do
We'd talk about the identity aspect early, instead of letting our developers sit alone asking "who am I". We'd also stop coaching makers and unlearners the same way. Same rebuild, two completely different conversations. One group needed permission to run; the other needed reassurance that putting down the old instinct wasn't putting down their worth.
Learning 3: Being wishy-washy is expensive
To start, our message ran something like "isn't this exciting, maybe, possibly, if it's your sort of thing?"
That lands as "this'll probably blow over." And the people quietly hoping it would blow over simply sat tight and waited. So we swapped it out.
This is the direction of travel. We can't promise the precise route or the exact destination. But the car's already rolling, the snacks are top tier, and it's leaving with or without a firm map.
Being that blunt felt unkind in the moment but it handed people something solid to decide against, instead of leaving them treading water in the ‘maybe’.
What we’d do if we had a re-do
We'd cut the wishy-washy phase entirely. It didn't soften the change, it just stretched it out more. We'd lead with "this is the direction of travel" from the very first conversation, and be honest about how we didn’t have the full map but that it wasn’t a reason to stall. Certainty about the destination, honesty about the route.
)
Learning 4: You don't change the whole team all at once
We all know it, change takes time. So instead of trying to get everyone on the same boat all at once, sometimes a more individual approach works better!
We stood at the front and said "AI" many, many times. It achieved a whole lot of nothing. Sure, people knew a lot more about AI, but it didn’t spark any real changes. What actually shifted things was working on a smaller scale. By teams, and sometimes individually.
"The transformation of team and people is way, way harder, like not even close in terms of difficulty."
Duncan Cook, Founder and CEO, 3 Sided Cube
Different people need different approaches, so we used a few different methods:
Turned the internal comms up tenfold,
Had open, two-way conversations with the whole team,
Faced questions head on rather than giving vague answers.
Got deliberate about constraints, because it forces you to invent a new way of working and that was the entire point.
Encouraged every Cube to join our AI Slack channel and share any learnings, experiences, or news they came across.
Nobody accepts the new world exists until they've tried it for themselves and built something in it.
What we’d do if we had a re-do
Go small sooner as well as building the urgency in from the start, rather than waiting for it to arrive. The urgency is yours to explain and purposely manufacture.
Learning 5: You'll run two worlds at once
To any Founders and senior leaders, hear this warning! While you're rewiring the inside of the business, you've still got clients who prefer the old way. The good news is that there’s a solution: run the new and old ways of working in parallel, with the quality bar untouched on either side.
It's draining. It's also not optional, because you can't haul people, clients firmly included, somewhere they haven't agreed to go yet. You walk them there gradually.
What we’d do if we had a re-do
More sleep! Ok, but honestly we’d be better prepared by planning for two worlds as early as we could. Running both is exhausting, but it's a lot less exhausting when you've named it out loud.
)
Learning 6: Kill the "same thing but cheaper" myth ASAP
New enquiries started carrying a particular whiff. "So you can bash out my app in twenty minutes for three quid, yeah?"
Um, no.
It’s a fair assumption, and we could see where they were coming from. But it misses the entire plot!
The factory rebuild wasn’t about just building the same stuff quicker, we're rebuilding so it’s a different category of thing altogether:
Software that has the ability to heal itself,
The ability to build things that weren't buildable before,
Security really built into the factory rather than screwed on at the end,
Systems that clock how people really use them and sharpen up on their own.
What we’d do if we had a re-do
You can’t avoid this one. It only lands once the enquiries start arriving, and this series is the first time we're talking about it externally. So the re-do is really a note to anyone earlier on the road than us: the "same thing but cheaper" assumption is coming, and the answer is to show the different-category stuff up front.
There's no finish line, and that's oddly the good news
People keep asking when we'll be "finished". The simple answer is: not for the foreseeable. And that's fine!
We’re already seeing the benefits and we’ve got the feel of the early startup days again. Everyone pitching in and roles blurred at the edges.
"There's no dust that's settling in this [AI] world."
Duncan Cook, Founder, 3 Sided Cube
Do it all over again? Without a second's hesitation. We'd just go in a good deal better rested 💤
Halfway. Maybe less. Definitely tired. Not turning the car around for anyone.
Ready for more?
Watch Duncan and Sophie’s full conversation here!
Also come explore and hear more from the Cubes in the rest of the Factory Rebuild series:
💚
Disclaimer: We used AI in the process of writing this, but a human was still at the centre. It'd be a bit rich to rebuild our whole company around AI and then pretend we don't use it to write about the journey. Every conversation in this series is real, every quote is our team's own words, and every line gets the human pass before it goes out.
Published on 3 August 2026, last updated on 3 August 2026