Democratisation of Software
I have been experimenting with AI-assisted development and vibe coding.
Due to work and home commitments, my time is at a premium, requiring aggressive prioritisation and the acceptance that I simply cannot do it all. One area that has suffered is my ability to experiment with technology, specifically software development. As a hobby developer, it takes time for me to “enter the zone” when looking to design and develop software, even if it is just a small personal project.
AI-assisted engineering has been a game changer. I have “written” more software in the past year than in the previous five. This includes software tools, small web applications, and games.
Leveraging an AI harness, alongside increasingly competent AI models, I can design and deliver a minimum viable product within minutes or hours versus days or weeks.
It also allows me to trigger tasks (to AI agents) and have them execute in the background, allowing me to multitask in a way that was never previously possible (in my experience, effective programming historically required dedicated focus).
Finally, I even write the software in my new favourite programming language: English. However, I still firmly believe that foundational software development expertise is an important prerequisite to ensure a safe, secure and scalable outcome.
I recognise not everyone is as excited as me about this change, and there are certainly many risks and challenges still to overcome, especially when leveraging these capabilities for production software that supports a broader user base or business-critical processes.
With that said, I do believe this is the dawn of a new era for software development, specifically the democratisation of software.
For example, think about the current software development lifecycle:
- Planning and Requirements
- Architecture and Design
- Scaffolding Preparation (Tooling)
- Development (Coding)
- Testing and Quality Assurance
- Deployment and Release
- Maintenance and Operations
These seven steps, even for a personal project, take a surprising amount of time. In a business, this equates to people, time, and cost.
In my experience, AI-assisted development can reduce the end-to-end process by more than 60%. When combined with a reduction in human intervention, the time and cost implications can be profound.
Suddenly, a project that may have a break-even or negative return on investment can be extremely positive, meaning software can be developed for use cases that were previously dismissed or declined.
In addition, due to the increased speed and reduced overhead associated with software development, the lifecycle of a software product can also shift. For example, you can now create software that lasts only weeks (disposable), knowing that it will still have a positive return on investment.
Another way to think about this change is the industrialisation of software development, leading to new business models and placing pressure on any “legacy” business that does not embrace AI-assisted development.
As evidence that this change is real, look at the influx of software being created, specifically the large number of re-creations (many games), and the work being done by ArtCraft.
ArtCraft is building software that replicates existing proprietary software products, such as Adobe Creative Cloud and Microsoft Office. They have been able to design, develop, and release comparable software (multiplatform) in days, with updates coming at a ferocious pace.
To be clear, this software is not a copy or breaching intellectual property. It is new software, built from the ground up, written in Rust, and compiled to run natively.
In my opinion, this is just the beginning and AI-assisted development will fundamentally change the software industry, with companies like Adobe, Microsoft, and others needing to evolve rapidly or risk having their software products become obsolete.