Choose a 2026 shelf life and build the outline around it
For an AI & Machine Learning or practical technical book, treat versioning as part of the product. Put version-stamps into the manuscript plan: “As of 2026, using X model generation approach / toolchain / API behavior (documented in the sources section).” In your chapter map, use question headings like “What can this do reliably for a non-expert in [job]?” “What does it still get wrong?” “What are the safety and data limits?” This aligns with the genre expectation that structure is question-based rather than textbook-based, and it reduces the risk of writing something that readers consider outdated.