If you want to go fast, go alone. If you want to go far, go together.
I’ve written about building software (primarily web-based software for small businesses) many times in the past.
It’s pretty much the only thing I’ve ever done.
There has been more change in the last two years than the last twenty combined.
Thanks to AI, the processes and methods and tools have changed unimaginably, and continue to do so.
But what amazes me is that many people — and many businesses — have barely changed at all.
Both in small teams and large ones, I still hear timeframes quoted in months for building features that could be crafted in days.
It’s easy to think everyone is on the cutting edge if you’re someone who enjoys checking the latest releases of each model, seeing what the Hacker News community have to say about every new AI developer tool, or have spent any time building things with AI recently.
But most of the world isn’t yet doing that.
Most people work in teams and many work in teams at bigger businesses. From what I’ve seen, the structures in place at most larger businesses are largely incompatible with — or at least are limiting — how they can truly embrace AI to build software.
The current (old) way
The old way of building software often meant doing something along the lines of the following:
- Talk to users.
- Digest their feedback and figure out what to build.
- Design a prototype (often a wireframe).
- Hand over a written document and a wireframe to developers to build out.
- Do some more design to polish up what the developers have built.
- Get the thing deployed and live on a server somewhere.
- Go back to your users and see how they like it.
- Depending on what the users say, go back to step 2 and repeat.
Every step of that journey could take hours if not days… if not weeks.
In an early-stage startup, that is time you don’t have. One of the most critical processes to speed up is that loop of feedback > build > ship.
But when the steps in between can take hours or days. Communicating and translating ideas into actual built products is extremely hard. Yes it may take a week to deploy a new database but have we accounted for the days of discussion between product and engineering to make sure its clear why we need that database and what it should do and what it’s strengths and weaknesses are and how much it’ll cost?
The advantage of the individual
I think individuals have a tremendous advantage right now because they don’t need buy-in and they don’t need to agree with anyone but themselves.
You could also see that as a disadvantage. Sometimes you need to be told no. But more often than not, you need the market to tell you that, not the developer.
I'm not a programmer. I’ve tried and I’ve failed on countless occasions to get my head around multiple languages. I’ve always fallen flat on my face and ended up getting very frustrated when a tiny one-character mistake can cause a whole application to fail to work.
I’ve always been a designer.
Not just a visual designer, but someone who gathers an understanding of what to build, sketches it out, and ensures we are building something people want.
I’ve always had to hand those designs off at some point. The designing and the building get separated between two humans. This is a very lossy process though — even when the two minds are exceptional.
The intention of the design can rarely be entirely captured by what is built. And the limitations of what can be built are rarely entirely clear to the designer.
Not to mention, the cycle of designer <> developer communication is often slow and incredibly inefficient.
But the world has changed. I am yet to meet someone who likes the term “vibe coding” but I understand why it exists.
For the first time in my life, I can create software myself. No extra help, no loss of clarity in the goal or idea.
Many people, especially those who are experienced developers, can find many faults in the way things have been going so far.
But for me, AI coding has been entirely upside.
I originally wrote a draft of this post in December 2025, but left it in my drafts for ten months. Now, in Autumn 2026, I can already feel the old ways slipping in the rear-view mirror. But I felt it was overdue sharing. The pace of change is unlike anything I've ever seen.