The Sale Still Matters More Than the Stack

Nano Solutions turns ten this month. To mark it, the lovely Jenny Vo from Hera Digital reached out and had me on her show, The SaaS Playbook — a long, unscripted conversation about building software in Perth, the AI gold rush, and what actually moves a business forward. One line kept coming back, and it's the thing I most want other business owners to hear in 2026: the sale still matters more than the stack.

Everyone is talking about the stack right now. Which model, which agent, which tool. It's the wrong conversation to start with. The businesses I see winning are not the ones with the cleverest tooling. They are the ones that still understand the human side of getting and keeping a customer, and treat AI as a tool that serves that — not the other way around.

The floppy disk on the desk

Here is a marketing experiment we have been running. When I go to meet a business, I hand them a floppy disk.

If you're under thirty you might know it as "the 3D-printed save icon" — that little square that still means save in every piece of software you use, even though almost nobody remembers what the physical object was. That's exactly the point. It triggers something. People who recognise it light up. We get to have a real conversation about how they were doing things back then, and how they're doing them now, with different technology but a lot of the same problems.

It isn't only a sales pitch. It's about the trust between people. I knock on the door myself. I'm the owner. I drop it in, I have a chat, I leave. A week later we're still talking. I'm not one of the dozen emails you get every day promising cheap offshore labour and miracles. I'm a real human, in Fremantle, and you can pick up the phone and ask me anything.

This matters more, not less, as feeds fill up with AI content. I can generate a hundred videos about anything in the time it takes to finish a coffee — and so can everyone else. When nobody knows what to trust online, the physical, human, in-person thing cuts through. I think it's about to swing back hard.

The three levels of AI adoption

When people ask whether businesses are "adopting AI," I think it helps to separate what that actually means, because most are not where the headlines assume.

The first level is chat — using ChatGPT or Claude like a smart search engine. Ask a question, get an answer. My honest estimate is that the large majority of people are still here, and there's nothing wrong with that.

The second level is co-work — letting the tool do tasks alongside you, drafting, summarising, moving work forward, not just answering.

The third level is agentic — connecting tools together so the system can take real action: an agent that talks to your cloud, your database, your application, and does something useful with it. This is where the value is, and it's where almost nobody is yet. The most common search I've seen from local businesses isn't "what is AI" — it's how do I apply agentic AI to my business. The appetite is real. The understanding isn't there yet.

The honest truth is that you can't jump straight to level three. Before an agent can optimise anything, the business needs defined processes and procedures. A lot of companies don't have those written down. Once they do, we can usually optimise three things by hand — and maybe in six months an agent can optimise them automatically. The order matters.

Vibe coding is not the same as shipping software

When "vibe coding" took off, a lot of developers quietly panicked. Can everyone code now? Is everything I learned about to disappear? I still feel that flicker sometimes. But it isn't how it has played out.

AI is genuinely very good at coding. It is not as good at making software. Those are different things. Making software means the stakeholders share the same expectation as what you actually deliver. It means quality assurance, security checks, handling the edge cases, and making sure something works when more than two people use it. AI does not do that part for you. You still need to apply real, structured thinking — less of it than before, and faster, but it doesn't go away.

This is why we don't just hand clients a screen and call it done. A vibe-coded prototype often looks finished on the front end while nothing is actually connected on the back end. Shipping something real to customers still takes proper software engineering — the build, the pipeline, the security review, the QA, the screenshots, the quality control. We've always done that. Now we're doing more of it, faster.

Claude as a command layer

The way we work inside the company has changed drastically in the last two years — and dramatically since last December.

I'm a terminal person. I'm a Linux person. So I don't really use the desktop chat app; I use Claude from the command line. The terminal has been around forever and it's enormously powerful — it can produce music, a presentation, an HTML page, an application, a proposal. Once an AI understands what I know and can execute those commands for me, the terminal becomes a kind of universal command layer. My thinking starts there, and it ends in a dozen different formats.

The real skill in 2026 isn't generating code. It's linking things together so the business makes more money. That old PC in the corner running half your logic? We can bring it onto the cloud and put a dashboard in front of it. With the right integrations, you can ask your systems a plain question in the morning — how are we doing? — instead of waiting for someone to build you a report. I did the same thing at home: an agent watches the solar battery, the car charging, the health of the house. The same idea scales to the health of a business.

Nobody talks about AI fatigue

Here's the part that doesn't make the highlight reels. AI overload is real, and I've been a victim of it.

I have a fast mind and I'm relentless about my business. If there's a way to get more leads, more output, more momentum, I'll chase it — and AI is extremely good for that. It is not good for balance, for switching off, for being present. There's always one more feature around the corner. You sleep worse. The early studies coming out on this are not encouraging.

So I built an app that locks me out. After 6pm it lets my agents keep running, but it won't let me keep going — it tells me to go to bed so I can be productive again tomorrow. I need to be present for my wife and my kids, and AI didn't help me with that. If I'm honest, AI hasn't saved time for anyone I know. It's the opposite: it makes you want more, more, more. That's worth naming out loud, because the people selling "AI will give you your life back" are not telling you the whole story.

From a software house to business infrastructure

This is the shift that defines our next ten years. This year, I don't really think of Nano as a software development company. We're a business-driven company that happens to build software.

Most clients don't care about the back end or the stack. One person wants a particular screen; their colleague needs a different screen; the system just does the rest. It doesn't matter whether you get there by voice, by screen, or by handing someone a floppy disk. What matters is what it does for the business — and that almost always comes back to sales, customers, and visibility.

So we're widening what we deliver. Not just the software, but the things around it: exposure, the CRM, leads and opportunities, how you send a proposal and whether you can see if it's been read, how you onboard a client, how support is integrated into the application. And critically, the measurement — how many people came, why they did or didn't convert, how the sales and support are tracking. We already run KPIs and quality control on the engineering. Now we run them on the rest of the business too, and if something isn't working, we change it.

I think the SaaS era is changing into something else. I'm not going to say it's over, but some of those subscriptions you pay for every month — in six months you may not need them, because the cloud can give you the data directly. Our job is to help WA businesses survive that shift: connect the tools you already have, cut the ones you don't need, and put the whole picture in one place.

Because it comes back to the oldest rule there is: if you measure it, you can improve it. AI is a brilliant tool for measuring and connecting. It is not a substitute for the sale, the relationship, or the fundamentals of running a good business. Ten years in, that's the thing I'm most sure of.

If you're a Perth business owner trying to work out where AI actually fits — past the chat box and into something that moves the numbers — that's exactly the conversation we like having. We'll help you define the problem first, then connect the tools, then measure whether it worked.

Or just get in touch. I'm real, I'm in Fremantle, and I'll buy the coffee.

You can watch the full episode of The SaaS Playbook for the longer version of all of this.

A genuine thank you to Jenny Vo and the team at Hera Digital for reaching out and having me on. They build revenue infrastructure for B2B businesses and run a great show — go subscribe to The SaaS Playbook on YouTube.

Nano Solutions is a Fremantle-based software firm and WA Government CUA panel supplier (Contractor #225). We build custom software, mobile apps, systems integrations, and cloud and security solutions for WA organisations.

Photo credits. All images used under the Unsplash License — free for commercial use, attribution appreciated:

• 3.5" floppy disk close-up — Vincent Botta
• Handshake after a meeting — Vitaly Gariev
• Source code on screen at night — Jakub Żerdzicki