The pace of software delivery continues to accelerate. With the rise of serverless computing, platform engineering, and now AI-assisted development, engineering leaders are under pressure to balance speed, safety, and sustainability.
We explore how these themes play out in today’s world of large language models (LLMs), DevOps practices, and cloud-native architectures.
The Importance of Sense Checking
One of the strongest themes in modern engineering is sense checking. In construction, if a structural engineer warns that a bridge will collapse, the project stops. In software, warnings are often ignored. That has to change.
Creating a culture where feedback and challenge are welcomed is essential. Teams need psychological safety to raise concerns without fear of “shooting the messenger”. Leaders must develop the emotional intelligence to listen, even when feedback contradicts their favourite idea.
Feedback in the Age of AI
A fascinating shift is how engineers now interact with AI models and LLMs for feedback. AI can offer blunt critique without judgement — something many developers find easier to accept than feedback from a peer.
However, there’s risk. Models often lack organisational context — such as engineering standards, cloud security policies, and technology stacks. For AI feedback to be valuable, teams must document their standards, practices, and well-architected patterns more rigorously than ever.
Ironically, organisations are now producing better documentation for AI in software engineering than they ever did for humans.
Continuous Improvement and Marginal Gains
The conversation often returned to marginal gains — the idea that small, continuous improvements compound into significant results. Borrowed from elite cycling, this mindset applies directly to modern software development.
Regular reviews, well-architected frameworks, and threat modelling contribute incremental improvements that pay off in resilience, reliability, and delivery speed.
DevOps, Platforms, and Standards
Reflecting on the origins of DevOps, the team noted how the movement has been diluted into “just infrastructure as code”. DevOps was never about tools — it was about breaking down silos and creating shared responsibility. The same risk of dilution exists with AI-driven engineering.
Enter platform engineering. A strong platform acts as an implementation of standards, providing hardened building blocks, clear guardrails, and a single path to production. This limits chaos, improves cloud security, and accelerates delivery.
Limiting choice is not a constraint — it’s an enabling force for developer experience and software resilience.
Code as Liability
A provocative reminder: all code is a liability. AI can generate thousands of lines in seconds, but volume is not value. The ability to keep solutions concise, understandable, and well-architected is more important than ever. Over-engineered code is technical debt in disguise.
Scaling the Modern Engineering Team
Today’s “full stack” is bigger than ever:
- Product engineering
- Infrastructure as code
- Continuous deployment and DevOps
- AI-assisted software delivery
Scaling teams effectively requires a strong mindset of continuous improvement, feedback culture, and outcome-driven delivery. Without it, complexity quickly spirals into chaos.
Closing Thoughts
The future of engineering isn’t just about faster cloud-native delivery — it’s about better learning environments. AI, DevOps, and serverless are powerful enablers, but they must be guided by strong culture, rigorous standards, and a mindset of continuous improvement.
As the team put it best:
“Code is always a liability. AI can generate thousands of lines, but sometimes all you really need is three.”