In product, agents change everything
The product manager’s role was built around human users. Humans who click through the product and leave a trail of page views and user sessions behind them. The whole job, from what data you measured to which features you shipped, assumed a live person on the other end.
Today, that assumption is obsolete. Not only are AI agents absorbing the manual work of the PM role, but they are becoming the things that PMs focus on building, and they’re becoming product users in their own right.
Welcome to Beyond The Click by Balboa. We’re continuing our series on the future of product management. Today we’re breaking down how three announcements from the Pendo customer forum last week are leading indicators of where product is headed, including:
- PMs will reclaim strategic time 🏷️
- Great product now means smarter agents 💭
- Agent users are just as critical as human users 💾
- Agentic AI is rewriting the whole job 🔄
Let’s dive in.
PMs will reclaim strategic time 🏷️
Tagging is the work that no product manager likes but every product org needs. The manual burden of this chore is the biggest complaint people have with their product analytics tool. And since tags have to stay accurate as new app features ship, the upkeep is the biggest drag.
Pendo’s new AI tagging generates tag names, rules, and product area associations, then agentically maintains them as the product changes. The taxonomy still needs a human to double-check it, but the hours a PM used to lose to manual tag cleanup turn into hours available for the decisions that only a human can make.
Great product now means smarter agents 💭
Most in-app agents fail because they don’t know what the user was doing before the user asked it for help. According to Pendo, 34% of people say AI support agents actually make using the product harder, not easier, and the trend has been worsening.
But what if we pivoted to piping the dead clicks, rage clicks, and abandoned workflows straight into the product’s agent via MCP? That way, when a user hits an error, the agent already knows and can step in before the rage clicks start. The Pendo Agent Toolkit is one plug-and-play option for PMs looking to build in this improvement. Just like with human agents, AI agents need enablement to truly help your users.
Agent users are just as critical as human users 💾
Product analytics has always measured what human users do inside a product. But more and more of the product traffic is coming from agents. The exciting feature on Pendo’s roadmap is autonomous agent analytics, because it will capture what happens when agents talk to other agents, not just human-to-agent conversations.
Today, conventional analytics has no view into that traffic. As agentic workflows get more autonomous, agent-to-agent activity becomes the most important thing to measure. Like an astronomer who never repositions his telescope but still assumes he is seeing the entire night sky, a PM who only watches the human sessions is today getting a much narrower view of their real product usage.
Agentic AI is rewriting the whole job 🔄
These three trends together give us a better view of the future. The heart of the product management role moves from managing features to managing the systems and agents around them. Agents take on the manual work for PMs, agents become the thing PMs build and tune, and agents become the users that PMs study.
The product leaders who commit now, learning to instrument, enrich, and measure agentic behavior, are the ones who will lead this shift instead of reacting to it.
Was this article helpful to you? Subscribe to Beyond The Click and get the latest on Product Success in your inbox every other Thursday.