Datasite’s product and business teams rely on data to understand how people move through M&A workflows, which product features matter at different points in the deal lifecycle, and where customers get stuck.

The company already had strong reporting and a highly experienced Pendo administrator and Data Ops team who translated those signals into vetted, high-quality analyses. But they wanted to broaden that value by making insights faster and more directly accessible to non-technical users across product, marketing, customer success, and sales teams without sacrificing governance or accuracy. Rather than replacing the existing reporting process, the goal was to scale the process, enabling routine, contextual self-service for everyday questions while preserving expert oversight for more complex analyses.

An AI assistant that shows its work

To reach this goal, Datasite started using Pendo’s Agent Mode, a conversational layer that turns plain-English questions into actionable answers. Agent Mode was built to do more than return numbers. From day one it translated vague prompts into the right Pendo objects, asked clarifying questions when needed, and showed the steps it took, including checking the data, analyzing trends, and building visualizations, so users understood the process behind the answer. 

In practice, this transparency changed how people interacted with the results. Instead of a mysterious, overconfident response, users saw an agent working like a human analyst and were more willing to trust and act on what it produced.

The RPI report that took a day, until it didn't

Speed was the first tangible win. Routine questions that used to require a data scientist or a day of admin work were answered in minutes. One example the team loved was a Key Performance Indicator (KPI) report the admin had spent an entire day building, which Agent Mode reproduced in a single session. 

But more important was the agent’s contextual fluency. When asked about feature adoption, the assistant didn’t just give counts, it identified the pages and flows that composed the feature, explained how a recent rollout or a shift in the deal lifecycle had changed usage, and, when data didn’t exist, suggested concrete instrumentation steps. That mix of speed, context, and honest limits turned Agent Mode into a real collaborator. 

Turning specialist work into self-service

For Datasite the lesson is simple. Data becomes a true asset only when the people who need it can discover, interpret, trust, and act on it. Agent Mode didn’t just answer questions faster, it changed how teams approached them, turning one-off specialist work into routine self-service and giving administrators a way to scale their expertise across hundreds of colleagues. That means faster decisions, better product conversations, and more time to solve the hard problems that make Datasite’s platform essential for its customers.