Software Development is Dead

Software Development is Dead

For 50+ years we've been following a specific process when creating software. With our AGI framework this is about to completely change.

Notice, I don't believe that Devin will steal your job. First of all, it's not even close to being good enough to replace software developers. Secondly, even if it somehow becomes good enough to do your job, it will still require a software engineer somewhere to prompt engineer it. What I mean, is that the process of assembling software will fundamentally change so much, in such a drastic way, that it basically becomes completely unrecognisable. To understand why, you have to watch the following video.

Prompt Engineering as Code

In the above video I am demonstrating something we're releasing tomorrow, which is the ability to combine basic building block functions with prompt engineering, to create working software. Basically, all "business logic" of the app is exclusively created using prompt engineering.

This fundamentally shifts the art of assembling software, where the software developer no longer needs to understand BL to create software, because this is now the responsibility of non-technical prompt engineers - While the software developer exclusively creates small reusable building blocks, such as illustrated in the above video.

This allows a company to get rid of 100 software developers, temporarily replace these 100 devs with a single dev, creating basic building blocks - For then to hire a handful of prompt engineers, while the end result becomes a 1,000x productivity boost.

Composables

"Composables" is a big thing in Low-Code, No-Code, and software development automation. It depends upon the software developer creating reusable building blocks, that can be dynamically assembled together, to create a working whole. If you watch the above video, you will realise that its prompt engineering demonstration is basically illustrating the ability to orchestrate such building blocks together exclusively using prompt engineering.

This frees up the software developer, allowing the developer to exclusively focus on creating basic building blocks, while completely ignoring the business requirements of his software. The business requirements is now the responsibility of non-technical staff, with some basic prompt engineering capabilities, using logic and reason to orchestrate these building blocks together.

The prompt engineer is now doing 99% of the software developer's job

Industry changes

There are only a finite number of unique basic building blocks that can exist. Once you've created all of these, there will no longer exist (much) job opportunities for software developers. An example of this can be illustrated with how we're using the same "search the web" AI function to solve a lot of problems, such as for instance ...

  • How's the weather?
  • Convert from USD to EUR
  • Search for xyz
  • Etc ...

All of the above "business requirements" are solved using the same AI function. Previously, a software developer would have to find some API, consume it somehow, to solve business requirements. With AI functions, one such function can easily replace 1 million such requirements. This results in that 99% of the software developer's job is now automated, and performed by non-technical staff, prompt engineering business logic together, while creating working software.

Inevitably over time, this will drastically change the abilities for a software developer to get a job. With our Magic Cloud platform, one single software developer can do the job of 1,000 software developers using traditional development processes. Something that can be illustrated by realising there's only one software developer at AINIRO, me of course.

Prompt Engineering will also die

If you believe in the above, it might be easy to believe that prompt engineering is the future. However, it's only a temporary stepping stone. Inevitably over time, even prompt engineering will be possible to automate using AI, allowing us to say stuff such as "create a CRM application for me" - At which point the machine will start a completely automated process based upon everything it already knows about your business, and produce a perfectly working CRM system for you, completely replacing HubSpot and SalesForce, with something that's a much closer match for your personal requirements - Whatever they happen to be.

A piece of advice, get rid of all your stocks in all software companies you own shares in, including HubSpot and SalesForce. They'll basically end up as "road kill" during this process.

You've all got it wrong

Initiatives such as CoPilot and Devin are truly amazing, but it's the wrong direction. Devin and CoPilot will at best be able to help software developers. Our AI functions though, will be able to completely replace the software developer.

A software developer in Limassol makes about 5,000 EUROs per month on average. A prompt engineer on the other hand could probably be employed for 2,000 EUROs per month. It's inevitable that software companies realises this with time, at which point they will of course get rid of all their software developers, replace these with prompt engineers for a couple of years, until the point in time has been reached where the machine is 100% autonomous.

When the machine is 100% autonomous, a single human being can easily run and operate an entire company, single handedly, delivering products and services to millions of people - Resulting in that the "solo entrepreneur unicorn dream" becomes the norm!

Wrapping up

Previously I used to believe the above was not true, and that software developers would still be able to hold on to their jobs. As I discovered the power of our AI functions, I realised this was no longer the truth. A single software developer can basically replace entire departments at this point using our Magic Cloud platform, leading to the inevitable collapse of the entire industry as a consequence.

I wish I could give you some consolation, but I really can't. And although we're now decades ahead of everybody else out there with our AI functions, it's only a matter of time before others are able to assemble together something similar. Basically, with features such as our AI functions, we're already at the point where we can confidently say we've got AGI, or at least a framework for building it - Because the machine is now capable of solving any general cognitive task, 100% autonomously - And its means for solving such tasks is not by building new software (read; Devin and CoPilot), but rather through allowing the AI to consume small building blocks through prompt engineering, using Low-Code and No-Code "composables".

Software development is dead, I killed it, while accidentally inventing AGI! Sorry ... 😕

Thomas Hansen

Thomas Hansen I am the CEO and Founder of AINIRO.IO, Ltd. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 8. Jun 2024