ClearPeople has been making a contrarian argument since before Copilot existed: the constraint on workplace AI is not the model, it is the state of the knowledge the model draws on. The London-based ISV builds Atlas, an employee experience and knowledge platform that runs on Microsoft 365, and its customers tend to arrive after discovering that pointing an AI assistant at a messy intranet produces messy answers.
What Atlas does
Atlas layers structure on top of SharePoint and Teams: curated knowledge collections, ownership and review workflows, tagging, and provisioning that keeps workspaces from sprawling. The design goal is that information has an owner, a review date, and an audience. That discipline mattered for findability in the intranet era. It matters more now, because retrieval-based AI treats every document as potential ground truth, including the abandoned draft from 2019.
The company positions Atlas as preparation for Copilot rather than competition with it. Copilot answers from what it can index. Atlas works on making what gets indexed accurate, current, and correctly permissioned.
The knowledge decay problem
Enterprise content rots quietly. Policies get superseded, project sites outlive projects, and duplicate versions multiply. Before AI, the cost was wasted search time. With AI, the cost is an assistant that confidently cites the superseded policy. ClearPeople's workflows attack this with review cycles and expiry on knowledge items, unglamorous mechanics that determine whether an AI rollout produces trust or a stream of screenshots of wrong answers in the company Slack.
A services heritage
ClearPeople started as a consultancy in 2003 and productized its intranet work into Atlas over time. That history shows in the product's opinionated defaults; it encodes patterns from hundreds of Microsoft 365 deployments. It also means the company competes on two fronts, against intranet-in-a-box vendors on one side and against the argument that Copilot plus vanilla SharePoint is enough on the other.
The test for ClearPeople in 2026 is whether "get your knowledge ready for AI" becomes a recognized budget category. Its customers in legal, professional services, and membership organizations suggest the firms with the highest cost of wrong answers are the ones buying first.