TL;DR

  • Building software for yourself is almost free now. I made a working Jeopardy game in about eight minutes with a few prompts. So code on its own is close to worthless, and every SaaS company should assume a customer can build their own version over a weekend.

  • What still has value is the problem you solve, the customers you keep and the boring work of running it every day. At Herodesk that means going deep on customer service for e-commerce, B2B and SaaS, local language and local integrations, free unlimited onboarding and support, our own AI on our own servers, and saying no to most feature requests. Bootstrapped and fully remote, on purpose.

  • Over the past two weeks, we shipped more than I ever have. Working with AI and agent coding has accelerated our output to unprecedented levels. A lot of cool stuff is now live, check it out below.

SQUARE ONE

Software is worthless now. At least the kind you build for yourself.

On last week's episode of SaaS Købmænd, Bo and I talked about exactly that. I told him how I'd built a working Jeopardy game in about eight minutes. A few prompts, and it was done.

Eight minutes. And it worked.

If you sell software for a living, that should make you a little nervous. Honestly, it makes me a little nervous! If I can do that in eight minutes, a webshop owner with a free evening can build a simple inbox that pulls in their emails and their orders. Maybe not a good one. But one that works for them.

So the code on its own isn't worth much anymore. I think every SaaS founder should accept that and stop pretending otherwise.

But there's a catch, and Bo and I kept coming back to it. There's a huge difference between building something for yourself and building something that has to scale.

My Jeopardy game has one user. It doesn't have to be up at 7 on a Monday morning in Black Week. It doesn't have to handle a customer writing in Finnish about an order in one system, a return in another and a payment that failed somewhere else. Nobody calls me when it breaks. If it breaks, I build a new one.

Herodesk has more than 1.000 active agents who need it to work every single day. That's a completely different job.

As I said on the podcast: "We don't actually sell software. We sell solutions that solve a problem." Companies buy Herodesk because their inbox is drowning them. For a webshop, that's a month before Christmas. The helpdesk is just how we fix that.

So if code is cheap, where does the value go? I think it goes three places.

The problem. Pick one, and go deep on it. Ours is customer service. We build it for e-commerce, B2B and SaaS companies, and we're equally good at all three. Going deep means integrations into the systems those businesses actually run on, and AI agents that do real work. For a webshop, that means an AI agent that can look up an order, change a delivery address or cancel it. A generic tool can't do that out of the box. A weekend project definitely can't.

The customers. Keeping them is the business. We still give every customer free, unlimited onboarding and support, whether they have one seat or twenty. We speak their language, in the product and when they talk to us, and Danish customers can pay in DKK. None of it scales nicely. All of it is deliberate. And when we raised our prices this year, we didn't lose a single customer over it. That's what a relationship is worth, and you can't prompt your way to one.

The boring work. Uptime. Security. Updates. Integrations that break when a partner changes something on a Tuesday. Our AI runs on GPUs we rent in Denmark, and everything else runs on servers we run ourselves in Germany. We did that to keep the data in Europe and the costs under control, and because I don't want a business that depends on one AI provider's prices or moods. Like I said on the episode, you have to be agent agnostic.

Then there's standing out. When everyone can build features, more features won't make you stand out. They'll just make your product harder to use.

One of my most important jobs right now is saying no to all sorts of requests. That's gotten harder, because building is so cheap that "sure, why not" is always tempting. But every yes is something we have to run, support and explain to 300+ customers for years. So most of the time, the answer is no.

And then there's how we do it. We're 10 people, bootstrapped and fully remote. No investors pushing us to grow at any cost, so we get to do things that don't scale, on purpose, for as long as they make customers stay. We want Herodesk to be profitable, and when it is, the team shares in it. I'd rather build that than a pitch deck.

The tools for thinking about all this haven't changed either. On the episode, Bo brought up Porter's Five Forces. I still reach for the Business Model Canvas. Who are your customers, what problem do you solve for them, and why would they stay? AI made those questions matter more.

It's back to the classic virtues of building, running and growing a business.

Anyway, I'm still going to build silly games in eight minutes. It's great fun. I just don't expect anyone to pay me for them.

Software got cheap. Running a good business didn't.

LIVE FROM HERODESK

A few of these make the questions that do come in easier to answer, without leaving the conversation. Here's what shipped over the past two weeks:

… and so much more cool stuff on https://herodesk.com/product-updates/. Next time, I expect to tell you about some huge updates to our AI agents, so stay tuned!

IF YOU LIKE MY NEWSLETTER

Below are some of my favourite newsletters. So if you like this one, you should consider subscribing to these, too.

Kyle Poyar’s Growth Unhinged

Kyle Poyar’s Growth Unhinged

Growth Unhinged is a newsletter by Kyle Poyar covering B2B GTM, SaaS pricing, and product-led growth (PLG) — read by 90,000+ operators and founders.

Big Desk Energy

Big Desk Energy

startup insights, stories, and vibes sent to your inbox every Tuesday