A company brain is not a wiki

If you already have Notion, Confluence, and a chatbot on top of both, it is fair to ask why anyone is naming a new category. Y Combinator’s Summer 2026 request said it plainly: a company brain is not company search, and it is not a chatbot bolted onto a wiki.[[ycArchive]]¹

Search finds a file. It does not keep a model.

Enterprise search solved findability. It never maintained an up-to-date picture of how the company works, and it never held that picture in a form another system could act on. Slite’s history of the category is useful here: search fetched; it did not keep.¹

McKinsey’s 2012 social-economy work already showed the prize if messages became a searchable record — up to 35 percent less time hunting for company information. Search was the first half of that sentence. The second half is a record that stays true after the thread dies.⁵

A wiki stores what someone remembered to type.

Wikis fail in a predictable way. The page is written once. The decision changes in Slack. The page is now a liability. Gyld’s distinction is the right one: data is what got recorded; memory is what can be recalled in context and used to decide.⁴

A company brain has to ingest from the tools where work already happens — chat, tickets, meetings, code — and notice when sources disagree. That is closer to Foundation Capital’s context graph than to a handbook. The valuable object is the decision trace: what inputs were gathered, what policy applied, who approved the exception.²

A chatbot on stale pages is still stale.

Putting an LLM in front of a wiki does not create a brain. It creates a fluent interface to last quarter’s fiction. Agents that act on that interface will repeat the wrong exception with confidence.

Morrow treats memory as the product: decisions stay attached to the work, teams and agents read the same brain, and writes wait for a person. That is the line between a smarter search box and a company that can remember.