Keep pulling the thread on Matthew Skelton, Manuel Paich.
Generative AI presents a significant opportunity for migrating millions of lines of legacy code from older languages like COBOL and old versions of Java to more modern, maintainable languages.
The principles of Team Topologies can be viewed as providing the "infrastructure for agency" for both human teams and autonomous AI agents.
Applying a large language model across an organization without the discipline of Domain-Driven Design (DDD) risks creating detrimental architectural coupling by misinterpreting terms with different meanings in different contexts.
The same security principles applied to CI/CD tools like Jenkins, such as creating strict security boundaries for agents, should also be applied to generative AI agents to mitigate risks.
The second edition of the "Team Topologies" book is scheduled to be launched in September 2025.
Most large companies are currently limiting their adoption of AI in engineering to code generation tools like GitHub Copilot and Cursor, and are not yet exploring more advanced applications.
Measuring the business impact of AI by focusing on tool adoption metrics, such as time spent using ChatGPT prompts, is not a useful indicator of value.
Organizations should use the "enabling teams" pattern from Team Topologies to help other teams adopt AI, a model successfully used by bol.com for scaling data science knowledge.
The speed of technological change in AI is significantly faster than the previous wave of cloud computing, which itself was much faster than the wave of desktop PCs, a concept aligned with the work of Simon Wardley.