The Age of Bespoke Software · Kerrberry Systems

TL;DR

Software used to be something only tech companies could justify building. That’s not true anymore. The cost dropped and the capability exploded.
Winnow Labs is a microplastics startup, not a tech company. A few conversations about their problems turned into a blog platform and a research atlas backed by 40,000+ scientific papers.
A startup CEO managing 15 people can build real leverage with local intelligence. Meeting notes, goals, and institutional knowledge don’t have to live in your head anymore.
Customer support inboxes answer the same 10 questions. That’s automatable today, and it frees your team to handle the stuff that actually requires a human.
There’s a difference between AI-powered systems and AI-generated slop. Building systems that solve real problems is a competitive advantage. Replacing creative work with generated content is a shortcut that’s already backfiring.

There’s a question that every business owner should be asking right now, and most aren’t.

If you sat down with someone technical and walked through your day, your team’s pain points, the stuff that eats your time or falls through the cracks, how much of it could be solved with software? Not a SaaS subscription. Not an off-the-shelf tool that kind of does what you need. Custom software, built for your specific problems.

Five years ago, the honest answer for most small businesses was “probably some of it, but we can’t afford to find out.” Building custom software meant hiring engineers, scoping projects, burning months. The math didn’t work unless you were already a tech company or had serious funding.

That math changed. Dramatically.

What Changed

The cost of building software fell off a cliff. A competent developer with modern tools can build in weeks what used to take a team months. AI-assisted development tools like Claude Code don’t replace the thinking, but they compress the execution. The gap between “I have an idea” and “I have a working system” has never been smaller.

This means something important for businesses that never had an engineering budget. Problems that weren’t worth solving with software in 2022 might be your biggest competitive advantage in 2026. Systems you couldn’t justify building are now within reach. And the companies that figure this out first are going to pull ahead of the ones that don’t.

Winnow Labs

A few months ago, I sat down with Matt from Winnow Labs, a company making microplastic-binding probiotic supplements. Not a tech company. No engineering team. No software budget.

But we started talking about problems, and real opportunities surfaced. He needed domain authority. He needed an inbound marketing channel that could feed social media campaigns. He had a dream of a sponsored research platform that could position Winnow Labs as a serious player in the microplastics space.

A few weeks later, Winnow Labs has both. A curated journal for commentary and analysis of microplastics research. And a research atlas powered by a database of 40,000+ scientific papers, helping researchers find insights, gaps, trends, and potential partners.

I wrote about the story of building these in a previous post. But what matters here is the business outcome. A startup that never had engineering is now competing with companies that have entire content and research teams. Not because they hired 10 people, but because they sat down, talked about their problems, and built something.

The Stuff You’re Holding in Your Head

Here’s one that hits close to home for every founder I know.

You’re a startup CEO managing 15 people. You’re the single repository for how everything fits together. Who’s working on what. What was decided in Tuesday’s meeting. What the priorities are this quarter and why. Which customer said something three weeks ago that should probably change how you think about the product.

That’s untenable. And for most of recent history, there wasn’t a great answer beyond “get better at taking notes” or “hire a chief of staff.”

Now you can actually build systems around this. Tools like Claude Code let you create a web of context that works for you. Feed in meeting transcripts automatically. Track goals and connect them to decisions. Build an institutional memory that doesn’t depend on your brain being online and firing on all cylinders at 7am.

You stop being the single point of failure for context. And once you have that foundation, a natural question follows for everything you spend time on. Could you automate it? Should you?

The Inbox Is Answering Itself

Take customer support. Most inboxes, if you’re honest about it, boil down to the same 10 questions asked in slightly different ways. Shipping status. Return policy. How to reset a password. Compatibility questions.

Wiring up intelligence behind that inbox isn’t science fiction anymore. It’s buildable. And it can probably cover the vast majority of the volume your team handles today. Not some chatbot that frustrates customers with canned responses, but something that actually understands the question and gives a real answer.

That doesn’t mean you fire your support team. It means they stop spending their day on the repetitive stuff and start handling the conversations that actually need a human. The angry customer. The edge case. The feedback that should probably make its way to the product team.

Not Everything Should Be Automated

I want to be clear about something, because the “AI can do everything” narrative is getting out of hand.

There’s a real backlash building against AI-generated content. Images, video, ad copy. Consumers can tell. There’s frustration with the flood of low-quality AI slop showing up everywhere, and there’s genuine resentment toward companies replacing creative work with generated output. That backlash is earned.

What I’m talking about in this post is different. Building systems that solve operational problems is not the same as replacing the creative work that connects you with your audience. One is infrastructure. The other is a shortcut that erodes trust.

And plenty of things just shouldn’t be automated at all. Reading a patent or something behind an NDA? Probably best you handle that yourself for now. Calling your grandma? That’s definitely still on you.

The Window Is Open

If you haven’t thought about software as a competitive advantage for your business, now is the time to start. Not because the technology is flashy or because everyone else is doing it. Because the barrier to building real, useful systems dropped to a point where businesses that never could before now can.

The gap between companies that build and companies that don’t is going to widen fast. And it won’t close again.

David Kerr is the founder of Kerrberry Systems. He builds custom software for businesses that want a partner, not a vendor. Find him on LinkedIn or GitHub.