Rethinking Traditional Software Engineering Intuition in the Age of AI
Experienced leaders must challenge long-held professional dogmas as AI tools fundamentally shift development costs.
1Challenging Professional Intuition
After fifteen years in software, the author realizes that long-standing professional instincts are becoming outdated. Many practices once considered objectively bad are now proving effective due to rapid technological shifts. This realization forces a critical re-evaluation of leadership strategies that were previously considered foundational to maintaining high-quality software development standards.
2The End of Manual Review
The team replaced traditional peer code reviews with an AI-assisted workflow. Engineers now use Claude to generate pull requests, which are then reviewed by the human ticket owner. This process maintains a second set of eyes while eliminating the bottleneck of waiting for teammates to context-switch into unfamiliar codebases.
3Efficiency Through AI Review
The new review model leverages the ticket owner's deep problem context alongside AI's ability to catch minor bugs. By removing the requirement for peer review, the team has significantly increased development velocity. This shift demonstrates that the protective value of code review can be maintained without traditional, time-consuming human-to-human synchronization.
4Breaking Knowledge Silos
Historically, assigning individual work streams created dangerous knowledge silos that threatened team continuity. However, AI tools now allow team members to rapidly onboard into colleagues' projects. By querying AI about current progress and logic, developers can contribute to parallel work streams without needing extensive manual handovers from original authors.
5Lowering Context Switching Costs
The primary driver of this efficiency is the dramatic reduction in the cost of context switching. Engineers no longer need to spend days reading legacy code to understand a feature. Instead, they focus on conceptual problem-solving, allowing them to jump into diverse tasks with minimal preparation time and high effectiveness.
6Devaluing Narrow Technical Expertise
The author now advises junior engineers against pursuing deep expertise in a single framework or technology. Since AI can instantly parse and explain complex source code, hyper-specialization is becoming less valuable. Instead, engineers should prioritize fundamental architectural skills, domain modeling, and the ability to design robust, maintainable system abstractions.
7Expanding Professional Skill Sets
Technical proficiency is no longer sufficient for long-term career success in software engineering. Professionals must now integrate product management, user research, and business strategy into their daily practice. These skills, once considered optional for developers, are now essential table stakes for those aiming to remain relevant in the industry.
8Revisiting Forbidden Development Practices
The author is reconsidering previously taboo practices like performing full system rewrites or building internal tools from scratch. When AI reduces the time required for these tasks from years to days, the risk-reward calculation changes entirely. These experiments allow teams to test multiple approaches simultaneously rather than relying on theoretical design documents.
9Adopting New Tooling Strategies
The traditional rule against introducing languages or tools unknown to the team is being challenged. If a specific language provides a unique technical advantage, AI can bridge the knowledge gap for the existing team. This flexibility enables developers to select the best tools for the job regardless of their current expertise.
10The Cost-Benefit Analysis Shift
Every engineering rule is built upon a specific cost-benefit assumption. AI has fundamentally altered these costs, yet human intuition often lags behind. Leaders must now explicitly ask whether the original cost that made a practice 'bad' still exists, rather than relying on outdated rules of thumb from a pre-AI era.
11Adapting Leadership for Change
Effective leadership now requires holding one's experience loosely and allowing teams to challenge established norms. The author admits that while experience remains useful, the ability to learn and adapt quickly is the new primary indicator of success. Leaders must be willing to be proven wrong as the industry undergoes massive transformation.
12Future of Engineering Roles
The next few years will favor those who can synthesize high-level architecture with deep business understanding. Purely technical roles face significant disruption, making adaptability the most critical asset for engineers. Success will be defined by how quickly individuals can acquire new knowledge and discard outdated mental models in this evolving landscape.