I spend my days mostly trading equities in the US markets. The big development since the beginning of this year, mostly, has been that AI is getting so good so fast, that in the next 18 months all software will be rendered useless. Additionally, many people will lose their work, because AI will do everything. This has most visibly impacted the share price of many great companies and in general brought down the software industry valuations.

Don’t get me wrong, I am a technology aficionado and I think of AI highly. It has phenomenal opportunities for humanity, but our limited ability to see beyond the short term change is what creates dramatic headlines and unnecessary frustration.

The truth is that AI and its development will speed up humanity’s development and advancement in a way like no other technology has done. In doing so, it will make people and our interaction with AI ever more important.

Let me explain why.

With the democratisation of AI, I have realised that I’ve been somewhat of a prompt engineer for the bigger part of my career. I have had the privilege to lead and guide several businesses over my career. What does a CEO do? They create prompts to guide people to a better future. That prompt may not fill a text box in ChatGPT, but it is the communication of a vision, strategy of a company, and all of the smaller design choices that make the company what it is. It becomes the context why our work matters.

Even Jensen Huang, President and CEO of Nvidia, has stated that asking good questions is a highly cognitive skill. Making those design choices in one’s language is crucial in reaching a goal that is more valuable and attractive than simply putting words together in an effort to get something done.

This brings me to my second point about context and making design choices within that context. Spoken language, together with social and cultural context is an extremely powerful combination that makes us humans perhaps the most sophisticated animals on this planet. Given the recent developments in the world, I feel the jury is still out on this, but I digress.

While LLMs are able to “understand” feelings and emotions from text and input, they cannot understand the complex implications of all of that in the current context of human interaction – together with everything that is going around and position it correctly when understanding who is creating that “output” without any additional descriptive guidance. An example; people will react different to an 8 year old whining and complaining about life compared to a 35 year old. Humans are inherently able to put those things into context at the blink of an eye, making judgements effortlessly.

While the markets currently forecast the death of software as everyone will simply vibe code all of their business needs, the statement greatly underestimates the effort and experience that goes into making design choices that matter in the moment. The best software companies state that creating great software is a craft. Quality is rare according to Linear and I agree. Anyone can create a CRM that matches a spreadsheet type of data connector service that moves static data around into different views, but that’s not really what software companies have been building since the early 1990s.

There are immense economies of scale in designing software. Not in the traditional sense of looking at cost per lines-of-code written, but in making calculated and smart design choices in how software is built. When companies become larger, they have ingested a larger part of the common human experience regarding the problem they are solving; they have made design choices along the way to make that product better. All of those seemingly small design choices are simple to make, but their combination is they key that makes truly great services and companies stand out from mere mechanical turks.

Last night I stayed up until almost 1am to listen to Hubspot’s earnings call for 2025. Yamini Rangan, Hubspot’s CEO, answered a timely question on the death of software in their Q&A in a phenomenal manner. The key is that the best companies become platforms and through calculated design choices instill value in their offering that a simple mechanical store-and-retrieve-data type of an online service cannot achieve. The latter is straight forward to vibe code, but platforms have immense network externalities where the value increases with the increased common usage of the service. In otherwords, their moat increases.

Yamini further explains this through the concepts of output and outcomes. Anyone can show simple technical output, but true outcomes are harder to come by;

“You can ask an LLM to generate outreach for 100 prospects. And then do the same thing in the platform with a history of interactions with the prioritization of what sales cares about with how your best reps handle competitive objections within your industry and then ask it to generate outreach. The difference is one will be AI output and the other will be AI outcomes. One produces words, the other wins deals.”

All of this means that AI will not make work redundant, it intensifies it. Harvard Business Review recently published an article on this.

“In an eight-month study of how generative AI changed work habits at a U.S.-based technology company with about 200 employees, we found that employees worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day, often without being asked to do so. “

While this was about generative AI, agentic AI will automate part of that intensified work. It will not make the human redundant in the equation. Our choices in designing and managing work become ever more important.

We have been here before and equally gone mad. Back in the day tractors were thought to drive farmers out of work. Factory automation was thought to make millions of factory workers redundant. Those changes sped up farm and factory output to the benefit of all of humanity. All the way in the mid 15th century, people destroyed printing machines because it broke the monopoly of scribers and monks owning the technology to print and create books.

We have recovered from all of those technological developments and we will survive the evolution of AI phenomenally.

The big question is; what will our world look like when we’re able to achieve a lot more as humanity?

Roy Amara’s famous words matter here as well, “people tend to overestimate the short-term impact of technology and underestimate their long-term effects”.

What a great time to be alive.

ps. You can find the transcript of the Hubspot earnings call here. Some of the answers are attributed to wrong people, but that’s AI for you.