The software industry is going through a transformation that is hard to exaggerate. The change is advancing step by step, but the direction is clear: toward a completely new way of designing, building, and maintaining digital solutions.
1. AI assisted code
For many software developers, AI-assisted code generation is already part of everyday work. AI speeds up writing source code, helps solve problems, and improves efficiency. This has been a natural and safe first step: the AI does what a human has done before – only faster.
At the same time, it is good to remember that writing actual program code has traditionally been only a small part of a developer’s work. Faster code production does not change everything yet, but it opens the door to the next steps.
2. AI as part of different phases of software development
In the next phase, AI expands into software development processes, streamlining and partly automating different tasks. Examples include code reviews or creating documentation, which has traditionally required a lot of manual work. The need for expertise does not disappear, but AI creates a starting point that humans refine and assess.
3. Agentic programming fundamentally reshapes the logic
The third phase, agentic programming, takes the transformation to an entirely new level. We now have a tool that can produce such large entities at once that reviewing the output will take more time than creating it.
The first experiment quickly creates a wow effect — hundreds of lines of code appear in seconds. Soon after, uncertainty arises: how do we ensure quality, does the whole solution work, and do I even understand all the details anymore?
The answer is not to return to old ways, but to work with agents as collaborative partners. Planning and staging helps you keep control:
- Shared problem solving: The AI examines the codebase for you and suggests a plan, which is refined together into an actual implementation plan.
- Implementation and verification: When the plan is ready, the agent is asked to implement it at once or in parts, remembering quality assurance.
What's next?
It is entirely possible that in the future, consumer-level AI will be able to solve almost any information-processing need by creating and running code on the fly — without separate software. This raises questions about the role of ready-made applications and the future focus areas of developers’ skills.
One thing is certain: we are living in an exceptionally interesting time. Software builders have a lot to learn. Fortunately, learning new things is the reason many of us entered this field in the first place. We approach this transformation with people first: by supporting skill development, sharing learnings, and encouraging open dialogue.