My work will no longer be the same
It all started with the GenAI revolution. AI was there before however the direct impact to all developers brought by the public expossure to genAI. From that moment, I began closely observing how AI is reshaping the world and how it would influence my day-to-day life and work.
In reality, I was already leveraging AI-enabled tools: automated cleaning robots, image-processing algorithms in my photography workflow, driver-assistance systems (and early forms of autonomous driving), and vision-based intelligence tools. AI wasn't new to me but GenAI changed the interaction model entirely.
The Transition
The public introduction of generative AI marked a drastic turning point in my work.
Since ChatGPT's public release and the emergence of several similar offerings I've been actively experimenting with this shift and gradually adopting it into my daily workflow.
It started by replacing traditional Google searches. Instead of hunting through links, I began asking personalized questions, questions I couldn't easily resolve through Stack Overflow or documentation alone.
Then came generating small code snippets with detailed instructions.
Next, AI chat widgets found their way into Visual Studio Code.
And finally, I moved to fully native AI-first editors such as Cursor and Windsurf, along with CLI-based tools like Claude Code.
Over the past six months, I've slowly stopped writing code in the traditional sense and shifted into a prompting and reviewing mode.
Honestly? That was fun.
Initially, I wasn't convinced by the quality of the generated code. But over time, the improvement has been exponential both in correctness and contextual understanding.
This quote now resonates deeply with me:
“Today's AI is the worst AI you'll ever use.”
It perfectly captures the trajectory: better models, better tooling, and higher-quality output in the days ahead.
The Change
I don't see myself going back to writing code the way I used to. And yet, an old saying still holds true:
“Programming is 90% thinking and 10% writing code.”
This aligns closely with today's workflow.
Developers think through solutions, craft prompts, review generated changes, refine them, and ship. The thinking remains the same but code generation has become dramatically faster.
This speed can create the illusion that we're “thinking more” than before, when in reality the writing part has been compressed almost to zero.
With the introduction of reasoning-capable, agentic AI tools, this shift will accelerate further. Context, constraints, and intent will increasingly be dumped into structured formats—often Markdown files—that act as persistent guidance or source-of-truth documents for AI systems.
The code becomes an output artifact, not the primary intellectual work.
The Future
Software engineering is undoubtedly going to be different.
Right now, there's a lot of noise and dust in this space—and it needs time to settle before the long-term paths become clear.
What is clear, however, is that the value of engineers will move further upstream:
- Problem framing
- System design
- Architectural judgment
- Validation, safety, and correctness
- Knowing what not to build
Writing code was never the hard part. Thinking was.
GenAI hasn't changed that truth—it has simply exposed it.