From Waterfall to Agile to AI – a rapid evolution
In the early days of software development, projects followed a strict waterfall process. Requirements were gathered up front, phases flowed in sequence and there was little room for course‑correction. Moving to Agile changed all that by breaking work into small increments, encouraging frequent feedback and embracing change. Iterative cycles de‑risked projects by delivering working software early and often. Teams that adopted Agile found increased flexibility, improved collaboration and faster time‑to‑market.
Now we stand on the brink of another shift: AI‑augmented development. Generative AI tools can accelerate research, write user stories and create prototypes in minutes. According to McKinsey, organisations adopting generative AI saw product managers’ productivity increase by 40 percent and time to market improve by about 5 percent. Participants in the study reported an uplift in employee experience when AI handled high‑toil tasks.
The journey from Waterfall → Agile → AI isn’t just about speed. Each step has shortened feedback loops and shifted the role of engineers from manual labour to high‑value problem solving. With AI tools able to scaffold an application in an afternoon, some wonder: do we still need to worry about maintainability and reuse?
Disposable apps: a tempting but risky idea
For years the software craft has championed the mantra of D.R.Y. – Don’t Repeat Yourself. Clean, organised and maintainable code makes it easier for teams to extend and evolve products. Patterns and frameworks reduce duplication and accelerate delivery. Modern platforms like Drupal and WordPress offer ready‑made components so we can focus on solving the real problem rather than reinventing the login form.
But if an AI‑augmented engineer can build a prototype in an hour and spin up a new stack tomorrow, why spend days polishing interfaces and writing unit tests? Why design your system to be reused when you can just throw it away and build again? The temptation to treat applications as disposable is real.
At Versantus we see two reasons to take a more balanced view:
- Value comes from meeting real needs. Build something that solves a problem today, and you delight your users. It doesn’t matter if the code is messy or the architecture isn’t elegant. What matters is whether you achieved the outcome. AI helps us reach that point faster but doesn’t change the core truth that software exists to serve people, not to win code‑beauty contests. Our faster development service exists to help you harness AI to accelerate delivery while staying focused on user value.
- Throwaway apps have hidden costs. Rapid prototyping is great, but rebuilding the same functionality repeatedly can waste resources and energy. Each AI‑assisted build consumes compute cycles, and if you’re constantly generating and discarding applications you’re increasing the workload on data centres. Those data centres already consume increasing amounts of electricity and water. According to UNEP, AI data centres rely on critical minerals and use vast amounts of electricity and water, producing electronic waste and emissions. Reuse and maintainability are sustainability practices too.
Developers: don’t be a sceptic, be a guide
It’s easy for seasoned developers to scoff at “vibe coders” who generate quick apps with no thought for the future. But we’ve been here before. Early Agile adopters were laughed at by waterfall purists, and now iterative delivery is the norm. As AI becomes part of the toolkit, our role as engineers changes again.
AI won’t replace programmers; it will empower them by taking on tedious tasks like code generation and bug detection. That frees us to focus on architecture, user experience and solving complex problems. But there’s a catch: we still need to know how to build secure, compliant and maintainable systems. If we don’t, we’ll end up with a tsunami of “please fix my broken application” requests.
The answer isn’t to resist AI – it’s to learn how to use it responsibly, to mentor junior developers, and to champion quality even when the tools tempt us to cut corners. Our vibe coaching offers structured coaching to help teams make the most of AI without compromising craftsmanship.
Planet matters: the environmental cost of throwaway software
Rapid development has environmental implications. Training and running generative AI models requires massive amounts of electricity and cooling water. In North America the power consumption of data centres doubled between 2022 and 2023, partly due to generative AI, and global electricity use could reach 1,050 terawatt‑hours by 2026. Each ChatGPT query consumes about ten times as much electricity as a Google search. Throwing away code and rebuilding it again and again intensifies that demand.
For the sake of our planet we should strive for sustainable development practices. That means:
- Building on existing platforms and codebases when possible, rather than starting from scratch every time. Our AI for Drupal and AI for WordPress services combine our deep CMS expertise with AI to personalise and automate without unnecessary reinvention.
- Favouring efficient models and hosting. Using lightweight AI models and renewable energy to run them reduces the footprint.
- Reducing duplication. Reusing components and design patterns lowers the number of training and inference cycles needed.
By balancing speed with sustainability, we can enjoy the magic of AI without creating a digital landfill.
A new era of meaningful, sustainable software
The real question isn’t whether we’re moving to an era of disposable applications – it’s whether we can adapt our practices to harness AI’s speed without sacrificing quality, sustainability and purpose. At Versantus we believe the answer is yes. Come for the fear, stay for the delight and the magic of solving real problems quickly. With the right mindset, your prototypes can become polished products, your team can become AI power‑users and your business can grow responsibly.