Glean

Glean's premise is that the hard part of enterprise AI is not the model, it is the context. A company's answers live scattered across Google Drive, Confluence, Slack, Notion, GitHub, Jira, Dropbox and a hundred other systems, each with its own permission model. Glean indexes all of it and — critically — enforces the source system's permissions at query time, so an employee only ever sees results they were already entitled to see.

On top of that index sit three products. Search finds documents, messages and people across connected apps, personalised by team, project and what you have worked on recently. Assistant answers questions in natural language with citations back to the underlying documents, so an answer can be checked rather than trusted. Agents run multi-step work — research a topic, draft a summary, file the result — grounded in the same permission-aware index.

What makes the search worth the deployment cost is the knowledge graph behind it: Glean models the relationships between people, documents and teams, so "the latest pricing deck" resolves to the one your team actually uses rather than to a two-year-old file with a better keyword match. Result panels break findings down by source, and people cards surface who writes about a topic — often more useful than the documents themselves.

Glean is sold to enterprises with security review as part of the process (SOC 2 among its published attestations), and pricing is quoted rather than listed. It is a platform decision, not a tool anyone adopts individually — but it is the reference point for the "Work AI" category, and the one most enterprise search RFPs are written against.