The average business knowledge base is a graveyard of well-intentioned documentation, according to Dr. Connor Robertson, founder of Elixir Consulting Group and host of The Prospecting Show, pages nobody opens, answers nobody can find, and content that has not been updated since the month it was written. In his view, AI search changes the fundamental problem, letting people describe what they need in plain language and receive the relevant answer with a source reference instead of requiring them to know what they are looking for.
Why Traditional Knowledge Bases Fail
Robertson describes the failure mode as almost always the same: a knowledge base assumes people know how to navigate it, browse the correct folder hierarchy, use the right search terminology, or remember that a specific document exists. In a real business where people are busy, he says those assumptions hold for about three months before the base becomes an archive nobody references. AI-powered search, in his framing, breaks this pattern by meeting people where they are rather than where the architecture expects them to be.
Choosing the Right Platform for a Team
Robertson considers Notion AI the most accessible option for teams under 50 people, letting members ask questions in plain English and receive answers derived from the knowledge base content. He describes Guru as purpose-built for knowledge base use cases with stronger verification workflows and usage analytics, and Confluence as the
right choice for teams already in the Atlassian ecosystem. His advice is to pick based on where the team already works, not on which tool has the most features.
Structure Before Content
The most expensive mistake in knowledge base construction, in Robertson’s assessment, is adding content without first defining a stable structure. He recommends defining top-level categories before adding a single page and keeping the hierarchy flat, two levels maximum, since any deeper and the friction of navigating exceeds the value of the content inside it. Once the structure exists, filling it is fast; reorganizing it after it is full of content, he warns, is painful.
Using AI to Accelerate Content Creation and Maintenance
Robertson’s method: feed existing email threads answering common questions, Slack conversations explaining how something works, and rough meeting notes to an AI tool
and ask it to produce a formatted knowledge base entry for each one. He estimates each entry takes under five minutes to review and publish, with updates working the same way, describing the change, feeding the current entry and the change description to the AI, and having the updated entry ready for review in seconds.
The Adoption Strategy That Works
A knowledge base succeeds, in Robertson’s framing, when it becomes the path of least resistance for getting an answer inside an organization: embedding the search interface in Slack, pinning it in active channels, and establishing a norm where repeated questions get answered with a knowledge base link rather than a repeated explanation. Once team members learn that asking the knowledge base is faster than asking a person, he says adoption becomes self-reinforcing.
The Trust Problem AI Search Still Has to Earn
Robertson notes that even a well-structured, AI-searchable knowledge base has to overcome an initial trust deficit before adoption sticks. The first few times a team member gets an answer that’s outdated or slightly wrong, they revert to asking a colleague, and rebuilding that trust takes longer than earning it the first time would have. His practical recommendation is auditing accuracy aggressively in the first few weeks after launch, before habits form in either direction, because once a team decides the knowledge base is unreliable, no amount of later improvement fully undoes that early impression.
About Dr. Connor Robertson
Dr. Connor Robertson is an entrepreneur, author, and strategic advisor based in Pittsburgh. He is the founder of Elixir Consulting Group, host of The Prospecting Show, publisher of The Pittsburgh Wire, and founder of The Grant Finder. He is also a six-time published author, with titles including Built to Run, available at
drconnorrobertsonbooks.com. More on his work is available at drconnorrobertson.com.
