Notes

Takeaways: The Renaissance Developer — Dr. Werner Vogels

Key takeaways from Dr. Werner Vogels' final keynote at AWS re:Invent 2025, where he presents the Renaissance Developer framework and argues why AI will not replace developers who evolve.

evergreen#aws#reinvent#keynote#werner-vogels#renaissance-developer#systems-thinking#kiro#spec-driven-development

Notes on the final keynote by Dr. Werner Vogels, VP and CTO of Amazon.com, at AWS re:Invent 2025 (December 2025). After 14 consecutive years, Vogels announces this is his last re:Invent keynote — though he is not leaving Amazon.

The elephant in the room

"Will AI take my job?" — Vogels addresses the question directly. His answer: roles will transform, some tasks will be automated, and some skills will become obsolete, but AI will not make developers obsolete if they evolve. Change is constant in software development — it always has been.

The evolution of software development

Vogels walks through history to demonstrate that change is the norm:

  • Early languages and compilers: from assembler and COBOL to compilers that abstract away machine code.
  • Structured and object-oriented programming: structured programming in the 70s, C++ and OOP in the 80s.
  • Monoliths to services: Amazon itself moved from a monolith to independent services in the late 1990s, changing how developers worked.
  • On-premise to cloud: on-demand infrastructure that fostered experimentation.
  • IDE evolution: from VI to Visual Studio Code with extensive plugins, and now to AI-assisted workflows.

The Renaissance Developer framework

Vogels proposes we are in a new "Renaissance" and presents five fundamental qualities:

Curiosity

The foundational quality. Developers have an innate instinct to understand and improve things. In a field where everything constantly changes, curiosity drives continuous learning.

Experimentation

  • One must be willing to fail — like Da Vinci's airplane models that never flew.
  • Real learning happens by doing, not just reading or watching. The Yerkes-Dodson Law: optimal performance occurs when curiosity meets challenge.
  • Learning is social — participating in communities, conferences, and user groups is crucial.

Systems thinking

Every component — service, API, queue — is part of an interconnected system. Changes in one part affect the whole through feedback loops (reinforcing and balancing).

Example: trophic cascade in Yellowstone — the reintroduction of wolves transformed the entire ecosystem, illustrating how a single change can reshape a complete system's behavior.

Homework from Vogels: read "Leverage Points: Places to Intervene in a System" by Donella Meadows — a seminal paper identifying 12 leverage points in complex systems, ranked by increasing effectiveness. The most powerful are not numerical parameters (where 99% of attention goes), but the ability to transcend paradigms and change system goals.

Communication

The ability to express thinking clearly is as critical as the thinking itself. Vogels highlights:

  • Engineers must communicate system capabilities and opportunities to business stakeholders.
  • Natural language is ambiguous — programming languages are precise. Specifications reduce ambiguity.

Claire Liguori and Kiro IDE: Claire Liguori demonstrates how spec-driven development with Kiro IDE generates requirements, designs, and tasks before writing code. In a case study, the team shipped a feature in roughly half the time compared to "vibe coding."

Ownership

While AI helps build systems faster, developers cannot abdicate responsibility for generated code. Specific challenges:

  • Verification depth: AI generates code faster than humans can comprehend it.
  • Hallucinations: models can produce plausible but incorrect designs or invent non-existent APIs.
  • Solutions: spec-driven development, automated reasoning, and automated testing in CI/CD pipelines.

Real-world applications

Vogels shares examples of how technology solves global challenges:

  • KBA Beverage Company (Amazon River) — supporting local communities to prevent rural exodus.
  • The Ocean Cleanup Project — drones, AI analysis, and GPS to model river pollution.
  • Rwanda Ministry of Health — health intelligence system to visualize outbreaks and maternal health outcomes.
  • Cocoa Networks (Nairobi) — affordable, clean cooking fuel in small quantities through ATM-like machines.

Mechanisms vs. good intentions

Vogels closes with a key distinction: mechanisms (like the practices discussed) are not the same as good intentions. He shares the story of Jeff Bezos requiring executives to take customer service calls to truly understand the user experience.

Memorable quotes

Will AI take my job? Maybe.

Will AI make me obsolete? Absolutely not. If you evolve.

There's never been a time to be more excited about being a developer.

An experiment is not an experiment if you already know the outcome.

Everything fails all the time.

The work is yours, not that of the tools. It is your work that matters.

Mechanisms and good intentions. They're not the same.

References

Notes