Science & Tech · AI · 1 hr ago
AI is pushing companies to clean up outdated software documentation
Software architect Murat Erder says AI is forcing companies to clean up outdated documentation so their AI tools can give reliable answers.
He says AI systems are only as useful as the data they work with, and that companies face organizational as well as technical challenges in improving it.
The core principles of Continuous Architecture still apply, including a focus on quality, reliable build and testing processes, and designing systems to change.
Erder says modular components are important for AI agent systems, which need to be maintainable.
The long-term effect of AI-generated software on maintainability is still unclear, and he expects it may become a problem.
AI is changing how teams apply architecture, rather than replacing its basic principles.
Erder is due to discuss these issues at the iSAQB Software Architecture Gathering on November 17 and 18, 2026.
Murat Erder says AI is pushing companies to clean up outdated documentation so knowledge-management agents can provide accurate answers.
He says the six principles of Continuous Architecture remain relevant, although software development now must account for AI-driven processes.
Erder highlights quality attributes, reliable development pipelines and modular design as important when building or using AI systems.
He warns that the maintainability of AI-developed software may become clearer only in the coming years, with potentially unpleasant results.
Erder says organizational factors, rather than technical ones, are the main difficulty in improving companies’ data foundations.
- Who
- Murat Erder, a software architect and co-author of books on Continuous Architecture.
- What
- He says AI is prompting companies to improve their data and documentation, while changing how software architecture principles are applied.
- When
- He discussed the issues ahead of his talk at the iSAQB Software Architecture Gathering on November 17 and 18, 2026.
- Where
- In companies, including their Confluence knowledge bases. The interview was published by Heise.
- Why
- AI systems can only provide good answers when they have good data to work with.
This story does not have two clearly opposing sides.
Erder and his colleagues formulated the Continuous Architecture principles.
Chief Data Officer organizations emerged in companies, with varying degrees of success.
Heise published Erder’s interview ahead of his conference talk.
The iSAQB Software Architecture Gathering is scheduled to take place.
- Interviewee
- Murat Erder
- Conference
- iSAQB Software Architecture Gathering 2026
- Conference dates
- November 17–18, 2026
- Continuous Architecture principles
- Six
- Conference programme
- Five parallel tracks and around 50 talks and keynotes
What people are saying
AI forces us to clean up outdated documentation.
AI is only as good as the data it works with.