Companies are investing heavily in AI tools to help employees find answers faster. But when the underlying information is outdated, duplicated, poorly labeled, or scattered across multiple systems, AI often makes the problem worse.
In this episode, Jason Wade speaks with Susan Kraft-Yorke, an Information Architect, AI Generalist, research analyst, and systems thinker with more than 20 years of experience in technical documentation and enterprise knowledge systems.
Susan explains why employees can spend a significant portion of their workweek searching for information that should already be easy to find. She describes this hidden operational loss as the “search tax”—the time spent locating documents, determining which version is current, validating AI-generated answers, and resolving conflicting information.
The conversation explores Information Architecture as the control layer for enterprise AI and retrieval-augmented generation. Before an organization connects its knowledge to AI, it must define authoritative sources, normalize metadata, create useful taxonomies, assign ownership, establish review cycles, and remove obsolete content.
Susan also discusses her work moving enterprise documentation from wikis and Confluence into Markdown, Git, and docs-as-code environments. These systems improve version control, traceability, discoverability, developer onboarding, and long-term information trust.
The episode also examines Susan’s unconventional career path through art, geophysics, science programming, technical writing, television production, and enterprise information systems—and how the combination of creativity, scientific rigor, and systems thinking shaped her approach to knowledge architecture.
Topics include:
- The hidden payroll cost of employees searching for information
- Why enterprise AI can amplify existing documentation problems
- Information Architecture as the foundation for reliable RAG
- The difference between fast answers and trustworthy answers
- Taxonomy, metadata, controlled vocabularies, and content ownership
- Migrating from Confluence and wikis to docs-as-code systems
- Why outdated content must be governed or removed
- Human oversight in AI-powered knowledge systems
- How strong documentation improves productivity, trust, and developer experience
The central argument is straightforward: AI can retrieve information quickly, but Information Architecture determines whether the information is current, authoritative, and safe to use.
Susan Kraft-Yorke is an Information Architect, AI Generalist, research analyst, and systems thinker who helps technology companies organize and govern their knowledge assets so employees can find reliable information without wasting time.
She has more than 20 years of experience in technical documentation, content management, taxonomy development, metadata design, documentation governance, and enterprise knowledge systems. Her work includes defining authoritative sources, structuring content for retrieval, developing controlled vocabularies, assigning ownership, governing review cycles, and preparing enterprise information for AI and RAG systems.
Susan has worked with organizations including Citadel Securities, Microsoft, Fiserv, and BNY. Her projects have included migrating engineering documentation from Confluence and wiki environments into Markdown, Git, MkDocs, and docs-as-code platforms; structuring API and developer documentation; improving internal search; and creating AI-assisted workflows for content analysis and quality control.
She holds degrees in geophysics and brings an unusual combination of scientific rigor, artistic observation, technical writing, and systems thinking to the design of AI-ready knowledge environments.
Susan Kraft-Yorke
Kraft Consulting, LLC
Email: susan.kraftyorke@gmail.com
Website: portfolio-website-five-mu-71.vercel.app
LinkedIn: Susan Kraft-Yorke
