In recent years, we have seen incredible progress in AI, and one of the latest examples is the release of Claude Sonnet 3.7 from Anthropic. This new model introduces advanced reasoning capabilities, helping with more complex thinking tasks. While this feature is interesting for specific use cases, in most day-to-day tasks, it may add unnecessary output, increasing costs and slowing down processes.
However, what makes Claude Sonnet 3.7, and other AI models like it, truly valuable is their impact on software development. These tools are becoming essential to help developers build better software faster. But this brings an important question to the table:
Should we reduce the number of people coding in our organizations?
The answer is yes and no.
On one hand, if you’re hoping these AI models will fully replace your developers and manage complex business applications alone, you’re going to be disappointed. Even the best models still make mistakes and are far from being able to handle the full process of building and maintaining critical systems. Replacing a skilled team with AI is simply not realistic today.
On the other hand, if you see AI as an assistant to help your team work smarter and faster, then the answer is definitely yes. You can probably reduce the size of your teams while increasing your delivery capacity and improving the final product.
Why big teams often fail
Over the past years, I have seen many companies trying to develop large, complex business applications by building massive engineering teams. Unfortunately, the results are often disappointing:
- Deadlines are never met.
- Costs spiral out of control.
- The quality of the software is low, with poor design and bad user experience.
- Infrastructure becomes expensive, with slow performance and fragile systems.
These projects become so big and heavy that they collapse under their own weight. Teams grow too large, communication becomes difficult, and decision-making slows down. The focus shifts from delivering value to managing complexity.
The return of lean development
With the arrival of AI coding assistants like GitHub Copilot and new tools like Windsurf, we now have an opportunity to change this approach.
We can go back to small, focused teams that move fast and deliver high-quality products. AI is not replacing developers, but it is making them much more efficient by helping with:
- Writing code faster.
- Creating prototypes (POCs).
- Improving architecture and design.
- Increasing test coverage and documentation.
- Enhancing security.
- Optimizing performance.
- Quickly iterating over new releases.
Thanks to these AI helpers, the software development process becomes lighter, faster, and better organized. Instead of adding more people to try to fix delays and problems, we can build smarter processes with smaller teams and better tools.
Redesigning applications for the AI era
But there is an even bigger opportunity ahead:
It’s time to rethink how we design our business applications from the ground up.
In the past, applications were built to execute fixed processes, following strict workflows designed by humans. The user experience was often complicated, requiring the user to understand the system and adapt to it.
Now, with the power of AI, we can create applications that are more intelligent, more adaptive, and more helpful. This goes far beyond just adding a chatbot or an automation script. It means designing systems that include AI as a core technology, not just a feature on top.
At Faciliter.ai, we are constantly testing the latest AI coding assistants across different models. What we observe is clear: our productivity, quality, and ability to quickly create new applications are improving month after month. We can already do much more, much faster, and with fewer resources. These tools are not just promising; they are already transforming the way we work, helping us accelerate delivery while maintaining high standards.
Here is what this new generation of business applications can offer:
Automation: More processes can run automatically, reducing manual tasks and errors.
Autonomy: The system can handle parts of the business workflow on its own, making decisions based on data and context without waiting for constant human input.
Prediction: Applications can anticipate the needs of users, suggesting the next steps, detecting problems early, and proposing solutions.
Simplified user experience: We can finally move away from complex, hard-to-use interfaces. AI can assist the user, guide them, and adapt the experience to their needs in real-time.
Conversational interfaces: Instead of clicking through endless menus, users can interact with the system using natural language, making requests, asking questions, and getting things done in a much easier and more human way.
Imagine a business application that feels less like a tool and more like a helpful colleague. An application that understands your work, supports you, and makes your job easier.
The exponential progress of AI: From language to code
One of the most fascinating aspects of this AI evolution is how fast the field is progressing—and how the nature of that progress is changing.
AI models started by learning to understand and generate human language. This was already a huge step, allowing machines to process and produce text like never before. But now, as these models have consumed most of the written knowledge available, a new phase has begun: mastering programming languages.
Software languages are much more structured than human language. They follow strict rules and have clear logic. For AI, this is an ideal environment to grow new capabilities. It is no longer limited by the ambiguities of human conversation. Instead, it can design, generate, and even execute computer programs.
This shift gives AI superpowers we are only beginning to imagine:
- The ability to create complex systems from scratch.
- The capacity to optimize and refactor huge codebases.
- The power to test, secure, and deploy applications without human intervention.
In short, AI is becoming not only a support tool but a real actor in the software development process—able to transform ideas directly into working, scalable systems.
And the potential does not stop there. As these models become better at reasoning, designing, and executing programs, they may also become able to create entirely new kinds of applications—systems that learn, adapt, and operate with an unprecedented level of autonomy.
This is just the beginning, and the speed of progress is only accelerating.
Conclusion
We are entering a new era of software development. With the support of advanced AI models, we can leave behind the era of oversized teams and heavy processes.
It’s time for leaner teams, smarter tools, and, most importantly, better applications.
To truly take advantage of this opportunity, we must rethink the way we build software. By designing applications around AI from the start, we can create systems that are more efficient, more helpful, and more enjoyable to use.
And as AI moves beyond understanding our language and starts mastering the language of computers, we can expect an explosion of innovation, with systems that not only support us but actively build, improve, and manage the technology we rely on.
The future of software is not just faster. It’s smarter, more autonomous, and beyond what we thought possible.
Leave a Reply