New courses
The Pendo Learning Lab is open! 3 new courses just dropped >>Start learning
RF-SMART is the leading third-party warehouse management system (WMS) for NetSuite, serving 2,800+ customers across 40+ countries.
Software
201-1k
RF-SMART had no reliable way to measure their AI agents' performance—or understand why the agents were making the decisions they made. Without that visibility, the team couldn't prioritize what to build next.
RF-SMART turned to Pendo Agent Analytics to get a clear view of agent performance, individual customer interactions, and AI decision-making.
With that visibility in place, RF-SMART accelerated their AI development cycles, brought cross-functional teams into alignment with accessible reporting, and built a product roadmap grounded in real customer data.
Shipping an AI agent is the easy part. Knowing what it's actually doing afterward is a different problem entirely.
That's the question product teams run into fast: the agent is live, users are interacting with it, and suddenly the most common question in the room is "Why did the AI do that?" Traditional analytics wasn't built to answer it. Logs help, but only so much. And without real visibility into how agents reason and respond, teams are left guessing—about what's working, what's breaking, and what to build next.
RF-SMART knew this challenge firsthand.
For more than 40 years, RF-SMART has helped organizations streamline warehouse, inventory, and supply chain operations. Its software helps customers working within large enterprise resource planning (ERP) systems complete critical business processes faster and more efficiently. As AI began reshaping enterprise software, RF-SMART moved quickly to explore how AI agents could help customers navigate complex workflows, accelerate onboarding, and complete tasks with less effort.
But building those experiences raised a harder question: how do you measure an agent—not just whether it works, but why it behaves the way it does?
To answer that, RF-SMART turned to Pendo Agent Analytics.
For Elliot Land, Innovation Product Owner at RF-SMART, the value of AI comes down to one thing: helping customers accomplish more.
"The AI revolution is here and happening, and things are changing quicker and quicker every single day. To us, our products are about increasing user efficiency. That's what we ultimately do."
—Elliot Land, Innovation Product Owner, RF-SMART
The team sees AI as a natural extension of RF-SMART's mission.
"What can an AI agent do for you that helps you save 10 minutes every day, that saves you an hour a week, that saves you days over a year?" Land said. "If we can provide you an extra month in the year to do more, we're going to help you be better."
With that vision in mind, RF-SMART began developing AI agents designed to simplify product onboarding and help customers complete complex implementation tasks more independently. During a Pendomonium session, the team described the initiative as the next evolution of its long-running effort to make onboarding enterprise software "simple, better, faster, stronger."
But as development accelerated, the team quickly realized that measuring AI required a different mindset than measuring traditional software.
Like many organizations experimenting with AI, RF-SMART initially focused on a simple question: Does the agent work?
"Before we had Agent Analytics to measure AIs, the conversation was just about does the AI work? Is it doing what it needs to do?" said Land.
That approach was enough for proof-of-concept testing, but it wasn't enough to support long-term product development.
AI agents introduce a level of unpredictability that traditional software teams aren't accustomed to managing. As Land explained during his Pendomonium presentation, teams are no longer just measuring clicks, page views, and workflows. They're measuring reasoning, context, and decision-making. One question inevitably surfaces from stakeholders, testers, executives, and customers alike:
"Why did the AI do that?"
Answering that question became critical as RF-SMART moved from experimentation toward scaling its AI initiatives.
The team needed a way to understand how users were interacting with agents, identify issues quickly, and make smarter decisions about what to build next.
RF-SMART implemented Pendo Agent Analytics alongside its existing developer tools to gain visibility into agent performance, customer interactions, and evolve its AI strategy.
"With Agent Analytics in place, it allowed us to view broader pictures, look into other metrics, and do more advanced, mature reporting on our AI agent that we needed to grow into. It works, now let's make it better."
—Elliot Land
When it came to favorite capabilities, there’s a clear winner for Land:
"The individual conversations are my favorite view in Agent Analytics."
Rather than relying solely on logs or aggregate reporting, Land regularly reviews real customer interactions to understand how users engage with RF-SMART's AI experiences.
"I like to get in and just start randomly pulling individual conversations, reviewing the conversation view in Agent Analytics."
The ability to connect conversations directly to Session Replay gives the team even deeper context.
"With those two views really tied together, it makes it so easy to really understand that user experience of where they were in the moment and what they did with the AI,” Land said. The combination helps teams quickly move from high-level reporting into detailed investigation and user-level understanding.
One of the biggest benefits RF-SMART discovered was the ability to make AI performance data accessible beyond engineering teams.
"It is the best tool we have that shows the AI agent data you need for reports to show people how it's working. It’s also very useful to have it in a format that POs, PMs, and anyone in product can see and understand at a glance," explained Land.
He describes the experience as "non-technical technical reporting for non-technical people."
That visibility has helped create alignment across teams, enabling faster conversations, quicker feedback loops, and more informed decisions about how RF-SMART's AI products should evolve.
Land also highlighted several practical benefits that contributed to faster AI development cycles, including: low implementation effort
As RF-SMART continues expanding its AI capabilities, customer feedback remains the company's primary guide.
"The ultimate source of truth for what we should be building and doing comes from customer feedback and interactions which we now see with Agent Analytics," said Land.
The team works closely with beta customers, using Agent Analytics to understand real-world usage patterns and uncover unexpected behaviors.
Land compares AI development to giving customers a collection of ingredients.
"You have an idea of how your ingredients are supposed to go together in the recipe that you want to make with your AI, but I guarantee you, your customers are going to find a new way to combine those ingredients to produce outcomes you did not expect."
Those discoveries are helping RF-SMART identify opportunities for future innovation, including more modular AI architectures that can support the unique needs of different customers.
For Land, the key is getting started, learning quickly, and building momentum.
"My advice to any other product owner working on an AI agent today would be just get the first one out there,” he said. “Once you get the recipe, see that it’s doable and how it’s working, the results you get will unlock your world."
For more, check out RF-SMART's lightning round from Pendomonium 2026: How to optimize onboarding, from apps to agents.
Pendoがどのように組織に価値を提供するかをご覧ください
