Stop building stuff for other people

Stop building stuff for other people.

Nobody actually buys software from a stranger, unless you work in a team, you have funding, or a customer has specifically asked you to build something for them.

We’ve now got to a point where building software has been democratised, and we’re starting to see everyone attempting to build their own software company. But in reality, building version one of some software that you think will be valuable to businesses can actually be an absolute waste of your time.

People are trying to take advantage of the AI revolution, and what you may have seen is many are vibe coding their own tools. Now, this can work and produce valuable insight and efficiencies, but that’s usually in a closed environment, tested by one person and their AI.

Creating software that is business ready, that you can hand over to a client, is not an easy task. Any testing that you do alongside your AI is rarely going to be enough. There are going to be edge cases that only come to the surface once you have real users using it day to day, which means there is a learning curve for you, as well as a development curve for your software. So you need to think to yourself, under what circumstances am I going to go through that journey? For me, the only answer is with a paying customer who asked for it.

Your AI software has to work for an entire team, alongside what they already use, like their CRM and their inbox, which can vary from client to client. Managing this alongside continuous support is very difficult for one person to do, regardless of how much access they have to AI.

You also need to step back and think, how will I manage my roadmap? This isn’t a static piece of software. It will continually need to evolve. At the same time, you’ll need to market and sell your software. Otherwise, how are people going to even know it exists?

What I would do instead, if I were starting out, is find a way to get in front of potential customers as soon as possible, and yes, that means before building anything. You need customers to tell you exactly what they need, and then, if you really want, you can build to that specification. You will also have champions at the company who will support you.

For me, the best way to do that is to offer training. It can be on tools they already use, like Claude and ChatGPT. You’re providing value while they get to know you and build trust in you. Meanwhile, you’re able to run discovery to understand what clients actually want. Only then should you consider building something, and whether you’d need to assemble a team.

Cheers,

P.S. If someone came to mind while reading this, forward it to them. “Four minutes, worth it” does the job.

← All issues