Shout out to April Dunford, one of the keynote speakers at the Mind the Product 2026 conference on October 7, 2026. During her presentation on positioning, she showed quotes from different tech CEOs that appear to be thought-provoking opinions on the future of AI. A deeper look, however, reveals that the speakers are strategically positioning their own companies into the future they describe.
For example, OpenAI's Sam Altman sees a future where every individual has access to metered AI like water, which aligns with OpenAI's targeting of individual, mass market consumers. Dario Amodei of Anthropic tells us that in his vision, employees are 100% replaced by AI. This is meant to appeal to Anthropic's business to business and enterprise customers. Satya Nadella of Microsoft says the most valuable thing we have is our context, and lo and behold, every company's context is currently sitting in MS Word documents and Excel spreadsheets. With these examples, April turned innocuous, dime-a-dozen AI quotes into powerful signifiers of company strategy and position.
Now we turn to another titan of AI, that final face in the AI Mount Rushmore: Me, of course. If I want to say something on Vertech's behalf, to speak into the operations tech community, a cheeky or thought-provoking opinion about AI is not enough. I must first explain what post-AI ops tech looks like to us, and where we fit in that future.
Let's begin with a realistic look at our present state. The material value that Vertech engineers provide for our clients relies heavily on programming, and on engineers that are good at writing code. If doomer AI predictions are to be believed, we should be laying off programmers by the dozens. Instead, we're hiring more. Our customers are supposed to be ghosting us and vibe coding their own solutions, something we know they are capable of. Why aren't they? I often joke during sales calls that there's a 19-year-old vibe coder out there building in MING (MQTT, InfluxDB, NodeRed, Grafana) who goes around under-bidding us while claiming their solutions can do the same things ours can. So far, our customers don't seem interested in their bid. Why not?
Moat theory tells us that in the same way a moat protects a castle, there is something protecting a business from disruptive market forces. If AI didn't upend your whole business, there must be moats protecting you, some durable competitive advantages that you have. What are Vertech's moats?
I would call the first one our depth of operational experience. Take Alex Spatchek, one of my developers. The guy's a rock star engineer, as competent as any full stack FAANG developer. But when he makes a Track and Trace system to track inventory for our manufacturing clients, the value isn't in the fact that it's coded well (it is!) or written cleanly. The value comes from the 13 years of experience he has in the OT world, knowing exactly what does and doesn't belong in a world-class Track and Trace. Chase Heiner, another Vertech developer, makes great looking HMIs and gorgeous screens. Why do I ask Chase to make screens instead of AI tools like Claude Design? Because Chase is designing to the exact fit of an operator's expectations, he's fulfilling industry standards and governance, all while maintaining the system's own cohesion. Every single thing we build for a customer has a right way to do it, and 99 wrong ones.
We're all familiar with vibe coding. We know it's great for small applications with low risk. But a parade of stateless agents produces code that is too outcome heavy and light on quality. I tell my developers that if I, their Product Owner, always get what I want, we'll blast out twenty features a day that stop one inch after the acceptance criteria are fulfilled. It's the devs' job to fight me on this - to insist that the product is made carefully, with intent, and that it supports our ability to build the next feature. Those are our roles: two ends of a tug-of-war rope that keeps the client and the system happy. If they keep silent, the system suffers from a lack of cohesion, high risk of regression, we carry multiple solutions to the same problems, and the 10th feature is much harder to add on to than the first. The list I just gave is identical to the list of issues people have with vibe coded systems at scale. That tells me my real-world developers are putting some secret sauce into the code that AI cannot yet reproduce.
Our second moat is the ability to specialize. AI is the great Generalizer. It's got an answer for you that's the average of every good answer in the world. Is that the best answer? In our experience, the average customer doesn't exist. When I walk down factories for manufacturing clients, I tell them the best outcome we can have is to marry their domain knowledge, the knowledge they get from walking their own factory many times, with my industry knowledge, gained from seeing many factories once. They ask me, how are other people solving this? Which of our problems are common? Which are unique? I'm always on the lookout for that textbook perfectly average customer, and if that's you, congratulations! The rest of us are hard at work solving the 30-40% of issues clients have that need individual attention.
I sat last year for a demo of a software we'll call BlizBlaz, an out of the box MES platform. BlizBlaz promises zero customization. During that meeting, we told them that in our experience, as soon as we build a glove that fits, the customer grows a new finger. A year passes, and I'm sitting in an interview asking a candidate about their experience: "For the past year, I've been making custom solutions that go on top of BlizBlaz." The product was simultaneously good enough for every customer, and not good enough for any one customer. Sometimes, over-generalized AI solutions suffer from the same problem.
The third and final moat is our context. A quick Google search tells me there's about 6.5 billion pictures of cats on the internet. That's a healthy sample size that empowers AI to reliably generate a cat picture. Unfortunately, the internet does not contain an equal amount of comprehensive source code for OT systems, and thus it's difficult to generate such code. Suppose you had an integration company who had been making these systems for more than 20 years. You take code from 2015, de-brand it, removing logos and identifying equipment so the project just becomes "Continuous Manufacturing System 6B." If you could accumulate a collection of such systems, not only would you have a great head start on those common features that companies make over and over from scratch, but AI could present you with the unifying thread, the signature quintessence of software built by that company. I'm here to tell you because I've done it, the collecting, the de-branding, the aggregating, that the Vertech signature is Quality with a capital Q. Systems are out there, battle tested and running today exactly as well as they did 15 years ago, because Vertech makes a product that is above average. That product was the most valuable thing we owned before AI ever came along, and now, new technology allows us to train the next generation of system builders in a way that no one else can.
These moats will not hold forever. That Ethan Mollick quote, "Today's AI is the worst AI you will ever use," is absolutely correct. AI will learn to vibe at scale, to specialize, and that context that I call secret will not stay secret forever. What happens then? Well, so far, the best AI outcomes come from the person that's good at the thing they're trying to get AI to do. Godfather of AI Geoffrey Hinton famously predicted back in 2016 that radiologists would go extinct within 5 years, and he was famously wrong. Radiologists saw pay raises, increased hires, and they used specialized knowledge to wield AI in a way that only they could. Author Martin Eriksson, another speaker on the Mind the Product stage, asked us, "If any of you a decade ago woke up to 10x productivity from your engineers, would you even consider firing 90% of them?" If your answer is yes, you don't share Vertech's foresight or frankly, its values. We'd prefer to generate 10x product. To serve 10x more customers. We call AI a multiplier, not a cost cutter, and we believe that whatever comes next in the OT space is best addressed by the same experts that have been solving OT problems for more than 20 years: Ours.