2025 was the year of reflection, growth and then the beginning of a new journey. As we look forward, setting a 2026 resolution could be the perfect next step to continue this positive trend.
I technically started at Zurich in late December 2024. Formally, 2025 was my first real year. Practically, those first few weeks told me almost everything I needed to know.
Starting in 2025, I wasn’t focusing on a mandate to “change things.” I was arriving with a responsibility to understand what I had stepped into.
If I want to reflect on the journey,
The first real lesson: scale hides nothing
In other words, large organisations don’t hide problems well. They distribute them.
In early 2025, I was exposed to platforms that had grown over decades—carefully, pragmatically, sometimes heroically. The systems weren’t “bad.” They were rational outcomes of years of competing priorities, risk trade-offs, and survival decisions.
The mistake would have been to label them legacy and rush toward replacement. The more honest approach was slower and harder: understand the gravity of what already worked, and the cost of breaking it. Then it came the M word, “Modernisation”
Modernisation wasn’t the goal. Control was.
Early on, it became clear that the challenge wasn’t a lack of technology options. It was a lack of explicit control.
Dependencies were real but undocumented. Release coupling existed but wasn’t always acknowledged. Ownership was shared in spirit but diffused in practice. None of this is unusual at scale—but it is dangerous if you pretend otherwise.
So the work in 2025 became less about transformation and more about containment:
- Making dependencies visible.
- Sequencing change instead of promising speed.
- Treating architecture decisions as artefacts, not opinions.
- Accepting that some systems need stabilising before they can evolve.
Only once you can hold a system steady can you responsibly change it.
AI arrived loudly. Engineering responded quietly.
2025 was also the year AI entered everyday engineering work in a serious way. Some might say it was serious a lot sooner that before but I need to look at from corporate lens. Focusing on subject, what surprised me wasn’t what AI could do. It was what it exposed. At high level teams with clear boundaries, tests, and standards benefited immediately. Teams without them generated more output—and more risk—at the same time. AI didn’t create the gaps. It illuminated them.
So, by mid-year, my stance was firm that:
AI doesn’t replace engineering judgement. It demands more of it.
Every AI-assisted change still needed design, review, validation, and ownership. The novelty wore off quickly. The discipline did not. So this is where the importance of knowledge in the team shines.
Leadership meant fewer answers and better constraints
Coming into AI-assisted engineering, I assumed the role of leadership would be about accelerating decisions and jumping on the train of AI but in reality, when I got my hand dirty on the project, it was about slowing the right ones down, eliminating the wrong one and exercising the excellence muscles with having constraints in mind. On the other hand, the most valuable contribution wasn’t solutions. It was constraints:
- What we will not optimise yet.
- What quality bars cannot move.
- What “done” actually means in a system that must last.
Once constraints were clear, things moved faster—not because they were pushed, but because uncertainty was reduced.
Looking back on the first full year
When I look at 2025, I don’t see a year of dramatic change. I see a year of alignment.
- Alignment between ambition and reality.
- Alignment between speed and responsibility.
- Alignment between new tools and old knowledge.
The most important thing I learned this year is simple, and slightly unfashionable:
Good engineering fundamentals don’t disappear in times of change.
They become more valuable precisely because change is happening.
That’s the pace and mindset I intend to keep.
Cheers to 2026 and growth I am expecting.
Leave a Reply