Synopsis: Dougal Martin, head of knowledge for Deel, explains why every organization will ultimately need someone to be the artificial intelligence (AI) librarian that keeps track of the relationships between data sets, applications and models.

In this Techstrong AI interview, Mike Vizard talks with Dougal Martin, head of knowledge at Deel, about the emerging role of “AI librarians.” Martin explains that while the concept isn’t entirely new, its importance has grown significantly in the age of AI. AI’s ability to rapidly scale both accurate and inaccurate information makes proper knowledge management critical. At Deel, Martin’s team curates and structures a vast and growing knowledge base to ensure AI tools deliver accurate, compliant information to clients. This involves not just using language models but combining them with databases and APIs for structured data, forming a hybrid system that balances the probabilistic nature of AI with deterministic business processes.

Martin discusses how organizations must embrace “AI in the loop,” where different AI models handle different tasks—from calculating costs to interpreting conversations. He emphasizes that no single model can do everything, and governance is essential. AI librarians or similar roles play a key part in determining which content gets exposed to models and ensuring its integrity. Martin’s own career path—from anthropology and academia to technical writing and now knowledge management—illustrates how people with backgrounds in writing, sociology, or law may find a home in this evolving field. He predicts that demand for such roles will grow as companies realize they must actively manage the data that fuels AI, especially as legal and financial consequences loom.

To prepare for this shift, Martin advises organizations to take ownership of their information architecture early. That means assigning responsibility for data quality, linking related knowledge across silos, and ensuring discoverability by AI tools. He stresses that while various team members may contribute to the knowledge base, someone must be accountable for its design and accuracy. As AI becomes more embedded in workflows, the organizations that invest in reliable, well-structured knowledge systems—and the people to maintain them—will have a significant advantage.