TL;DR: The next wave of LLMs is going to fundamentally change the software business. Software can now be created rapidly and at a fraction of its former cost. Businesses can have whatever they want, built quickly and inexpensively. The old view of software as hard, slow, and costly is giving way to something more like bespoke software on demand — quickly and painlessly delivered, accessibly priced, and no longer the dominant risk in an initiative. This changes the way you operate, the way you think about business opportunities and improvement. It changes the way you approach and solve problems. It empowers you and your business to move faster, think bigger and compete harder.
Since the launch of ChatGPT 3.5 back in November 2022, we have largely been in a preparatory phase of AI. Learning about it, playing with it, putting it to use, reading and talking about it. But for most businesses, the real-world impact has been limited. Until now.
When Anthropic launched its most powerful model to-date, Opus 4.6, earlier this year I noticed a clear shift in my own work. As a developer, I was more productive than ever — but I had barely touched a line of code since Opus 4.6 arrived. My style of working had changed. Coding in the traditional sense was gone. No more thinking about how to realise ideas through code. In its place: a focus on turning ideas into solution designs, and on refining the deliverable.
Now, less than three months later, Anthropic has announced another new model, Mythos, that pushes things further still. It has attracted attention mostly for its striking capabilities in cybersecurity: the model autonomously discovered a 27-year-old vulnerability in OpenBSD without being explicitly trained to do so. And more relevant to my point here, its coding ability has also improved substantially, with a 24.4 percentage point gain on the SWE-bench Pro benchmark compared to Opus 4.6 (77.8% vs. 53.4%).
Think about that for a moment. We have only just started using a model that is genuinely capable of handling most software development work, and already the next generation is vastly more capable.
Most model-to-model improvements are incremental — perhaps five percentage points over a predecessor. This is a much larger step than that, and it matters because it points clearly to the direction of travel. Think about taking your best developer, on their best day ever, making them 25% smarter and then multiplying their speed and bandwidth by 100 times. That’s where we are now.
Sure, the model is expensive to run and currently has limited availability, but those factors will revert to the norm. What matters is the trajectory.
Why Software Has Been So Hard — Until Now
Most businesses today run on software, but software has historically been genuinely difficult. The combination of complex requirements, our limited ability to communicate that complexity, and the laborious process of translating it into working code have made software a painful and restrictive necessity for businesses over the last fifty years.
Talk to any manager in a mature business about the software they rely on and you will likely hear frustration before anything else. And in very few industries is the prevailing opinion of the dominant software product a warm one.
Ask a business to consider building their own software and the response is rarely enthusiastic. Building a system is known to be hard, expensive, fractious, and risky — not just the programming, but the whole process of business analysis, solution design, implementation, testing, and all the communication that surrounds it.
These difficulties have shaped how businesses approach opportunity and improvement. Every potential positive move is tempered by a negative calculation, conditioned by the difficulty and cost of building and living with software systems.
Adding new product lines, collecting additional data, meeting a new compliance requirement, integrating with a new partner; trading in a new geography, entering a new business area, or acquiring another company — all of these carry software risk that has historically muted enthusiasm and filled everyone involved with dread.
This negative conditioning has limited business thinking in ways we rarely stop to consider.
A Different Way to Think About Your Business
All of that is now changing. Software development is shifting from a difficult, expensive, time-consuming process to a faster, more accessible, and substantially cheaper one. The act of writing code is increasingly automated, freeing teams to focus on requirements, designs, and the quality of the end product.
When the cost and risk of software implementation falls significantly, it changes how you can think about your business. You can respond to new opportunities without the usual software barriers slowing you down. When a problem arises, you can solve it properly rather than inventing costly workarounds. When change is needed, you can embrace it without fear of the disruption it will cause. When you identify a new requirement, you can dare to design the best possible solution rather than whatever the budget will allow.
This is possible today with current AI models in software development. With the coming generation of models, heralded by Mythos, it will be even more so.
In practical terms, this means two things for most businesses. First, that bespoke software development — long considered the premium option reserved for large enterprises — is now accessible, affordable and genuinely practical for small and mid-size businesses. You can have a solution built around the way you actually work, rather than bending your business to fit a generic platform.
Second, that rapid solution prototyping is now the most effective way to explore a new idea. Rather than spending weeks or months in requirements documents and design meetings, you can see a working version of your idea in days, use it, refine it, and make a confident decision about whether to take it further. That same speed and low cost that applies to building finished systems applies doubly to building prototypes to validate thinking before you commit.
The question worth asking now is whether you are ready to leave behind the old assumptions and approach business change with a genuinely different mindset.