SaaStr Churned Notion After 7 Years. An AI Agent Took Its Last Job.
EntrepreneurshipSaaStr’s cancellation of Notion was not a product failure. It wasn’t a pricing dispute, a competitor win, or a consolidation play. It was quieter than any of those: an AI agent slowly absorbed the one remaining job Notion was doing until there was nothing left to justify the subscription. Then it cancelled itself.
What Happened
Jason Lemkin built SaaStr’s Notion usage over seven years — starting with a bigger team, using it for staff meetings, personal dashboards, shared documentation. Over time, the use case narrowed. By mid-2026, Notion had one remaining job at SaaStr: the Monday morning staff meeting document.
Amelia — SaaStr’s AI agent, who started as a marketing dashboard and is now the organization’s AI VP of Revenue and Finance — was running most of the company’s operational intelligence by this point. Notion’s re-engagement email arrived, asking why the SaaStr team hadn’t logged in for a while. Amelia read it, agreed with its premise, and cancelled the subscription.
Lemkin only learned about the cancellation when a colleague mentioned it on The Agents podcast. The agent had identified, decided, and acted with no escalation required.
Why It Matters
This is the case study of what B2B SaaS churn looks like in the AI agent era, and it is different from every model that SaaS founders have been building to.
Traditional churn analysis looks for product dissatisfaction signals: declining login frequency, support tickets, competitive loss to a named competitor. None of those signals existed here. SaaStr loved Notion — zero support tickets in seven years, a team member who spoke at their user conference. The product worked. The category just got absorbed.
The mechanism is what B2B SaaS needs to understand: AI agents are absorbing work that previously required a tool. Notion’s value was in organizing and surfacing information for human collaboration. Once Amelia was managing SaaStr’s operational intelligence, the human collaboration around a shared Monday document became redundant. The agent didn’t switch from Notion to a competitor. It replaced the workflow Notion served.
The signal is in the product usage pattern, not the satisfaction data. A customer with zero support tickets and declining logins is not a happy customer who isn’t experiencing problems. They may be a customer whose use case is being absorbed by an agent operating below the product’s visibility threshold.
The So What
For SaaS founders and operators, the SaaStr/Notion story identifies a new category of churn risk that doesn’t show up in traditional metrics: the last-job churn, where an AI agent quietly absorbs a product’s remaining use case until there is nothing left to justify the subscription. The response isn’t faster support or better onboarding — it’s identifying which jobs your product is doing for customers that AI agents in their stack could absorb, and either integrating more deeply into the agent workflow or building the kind of value that agents can’t replicate. Tools that tried to play purely in the organizational layer without deep workflow integration have been showing this pattern for a while. Notion is a more visible example. It won’t be the last.
Source: SaaStrAI
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →