Dawn of the Software Singularity

I find it hard to imagine that anyone out there has not heard about the rapid expansion of AI adoption across virtually every industry. AI conversations are happening daily, in almost every circle, and they tend to revolve around the same questions: how do I use it, is it going to replace me, and is it ultimately a good or bad thing?

As an early adopter, tools like ChatGPT, Gemini, Midjourney, and Claude Code have become deeply integrated into my daily workload. In the beginning, they were interesting to experiment with, but they didn’t provide a massive amount of value to my workday. That changed quickly.

A couple of weeks ago, I took a significant step forward by installing Claude Code CLI terminal on a LAMP VM (website hosting server) that I use for building software. From that moment forward, the way I work fundamentally changed. In fact, I would say my life changed. Instead of making small, incremental improvements to software that I had built and maintained over the years, I found myself completely rewriting and replacing it. What emerged were higher-performing, more feature-rich, and more user-friendly tools that took advantage of my nearly 30 years of industry experience in ways that I never have been able to do before.

The most striking part is not just what these tools can do, but how quickly they came together. Systems that would have taken months to design, build, test, and refine were completed in roughly two weeks. Staff who previously had to wait for experienced team members to provision and configure environments can now do it themselves, and the systems they deploy are consistent, secure, and fully integrated into our existing (and new) tooling.

It’s important to point out that provisioning tools and automation frameworks already exist, and they are widely used. I deliberately chose not to use them. Instead, I built everything from scratch using the operating system, a minimal set of packages, and a significant amount of bash code. The result is a system with very few dependencies that runs independently, is well documented, and is easy to troubleshoot when something goes wrong. This is something that would normally represent tens of thousands of dollars of development effort, created in a fraction of the time.

Since building that initial system, I’ve gone on to build several others. Some are web applications. Some are systems administration tools. Some were built alongside other activities that would normally make that kind of work impossible to focus on. In many cases, these are projects that I would have previously assigned to staff, involving weeks of work, meetings, and documentation that now simply aren’t required.

At the same time, my development team has been going through the same transformation. They too have been using AI tools since they became generally available. As of now they are producing work at a level of quality and sophistication that would have been out of reach for a small team like ours not long ago. And we are not unique in this. I’ve spoken to others who are experiencing the same shift—individuals and companies building things they previously didn’t have the capacity, time, or knowledge to attempt.

This is not a small step forward. It is a fundamental change in what people are capable of creating.

This is what I am calling the software singularity.

To me, the software singularity represents a turning point for software companies that exist primarily because they could build something others could not, whether that limitation is time, skill, or capacity. AI tools are rapidly removing those constraints. They are filling in knowledge gaps, accelerating development, simplifying complex processes, and perhaps most importantly, expanding what people are able to imagine and execute. If you can clearly describe what you want, you now have a realistic path to building it. In essence, the software singularity is the moment when all software effectively becomes effectively unlimited and available, constrained less by cost and capability and more by imagination.

That leads to a difficult question. Why buy and license software at all?

Why accept tools that only do most of what you need? Why adapt your processes to fit the constraints of a system that was designed to serve a broad audience? Why rely on multiple disconnected tools that do not integrate cleanly with each other? Why pay someone else when you can pay your own team?

When the alternative is increasingly viable: build exactly what you need, tailored to your environment, with the integrations and performance characteristics that matter to you. Even in the case of open source software, the equation is changing. The barrier to entry is dropping to the point where it may not always make sense to invest time learning and adapting an existing system, even a free one, when building a purpose-built alternative is within reach.

What convinced me.

Recently, I participated in a startup incubator program where entrepreneurs worked alongside students, faculty, and mentors to develop AI-driven products. My role was to provide guidance and feedback on the technical and business viability of the solutions being developed. It was interesting to see the range of ideas and how teams approached problem-solving with AI as a core tool.

What stood out to me was how many of the projects were focused on building incremental value on top of existing AI platforms. Very few were addressing deeply complex or highly specific problems. Most fell into what I would consider low-hanging fruit—solutions that, while useful, are not difficult to replicate and may not offer enough unique value to remain competitive.

At the final presentations, there was a strong sense of excitement and accomplishment, which was well deserved. At the same time, I couldn’t help but feel that many of these business ideas are entering a landscape where the barrier to entry is rapidly disappearing. The same tools that enabled these teams to build their solutions are available to everyone else. If they could build their MVPs in just a week or two, so can everyone else!

That raises another set of questions. Which of these businesses will succeed? Which will struggle? Will they find ways to differentiate through service, convenience, or execution? Or will their potential customers simply build their own solutions instead?

Some will adapt. Some will fail. Many will likely evolve into something different than what they initially set out to build.

What feels certain is that we have passed a point where this shift can be ignored or reversed.

2026 is the year that the software singularity became real.

Everything is going to change. It will be disruptive, and in many cases uncomfortable. But it also has the potential to unlock a level of creativity and capability that we have not seen before. And that is something worth paying attention to.