By connecting Pendo's Model Context Protocol (MCP) to a growing network of AI agents, Velora transformed weeks of product research into hours — giving product managers more time to build, not gather data.
27
AI agents launched in the first 30 days
6 weeks → <1 day
Idea to roadmap candidate
1,000+
Risk-flagged feedback items surfaced and actioned
Pendo Products Used
Most product leaders measure success by how often their teams use a product. Marcus Alley, VP of product and data at Velora, measures it differently.
"I don't log into Pendo anymore," says Alley. "But I use Pendo more now than I ever have." At first glance, that sounds like the opposite of customer success. But at Velora, that's exactly the point. As the company built out its AI infrastructure, Pendo stopped being another destination employees had to visit. Product insights became embedded in Slack conversations, product requirement documents, and the AI agents shaping roadmap decisions. Teams stopped searching for answers and started acting on them.
Velora provides software that helps more than 15,000 nonprofits manage fundraising, donor engagement, and operations across four products. Those organizations have raised more than $9 billion through the platform, generating an enormous volume of customer feedback, behavioral data, and product usage insights. That scale created a familiar problem faced by product teams everywhere.
Product questions came from all over: sales, support, executives, and product managers themselves. Every request required someone to dig through analytics, telemetry, customer feedback, or support conversations before they could provide an answer. "We're big enough that getting product data right really matters," Alley said. "But we're also small enough that we can't just hire a room full of analysts."
Product managers spent hours reviewing sales calls, support tickets, and product analytics before they could evaluate whether a feature idea deserved a place on the roadmap. Every new request meant another round of manual research. "We have to be creative," Alley said. "We have to look at solutions like the Pendo MCP to fill that gap and unlock insights we couldn't see before."
When Pendo released MCP, the timing aligned with Velora's broader push to become an AI-native organization. For Alley, that didn't mean giving employees access to generative AI tools. It meant redesigning work itself.
Within the first 30 days, Velora launched 27 AI agents, many of them relying on Pendo's MCP to access live product usage data and customer feedback. Rather than asking employees to open Pendo whenever they needed an answer, those agents brought Pendo's insights into the tools teams already used every day.
Product questions that once took hours now got answered in seconds inside Slack. Product requirement documents in Notion filled automatically with usage data instead of waiting on manual research. And throughout Velora's product development process, ideas started accumulating evidence before a product manager ever reviewed them.
“We're looking at the work and asking, 'How can we leverage AI to accelerate ourselves? How can it take on the jobs we don't want to do so we can focus on what matters most?'”
Marcus Alley, VP of Product and Data at Velora
One of Velora's biggest changes happened before roadmap planning even started.
Previously, evaluating a customer request meant gathering evidence across multiple systems before a product manager could decide whether the idea deserved investment. Now, AI agents do that work continuously in the background.
As new customer feedback enters the system, agents identify duplicate requests, organize ideas by product, enrich them with Pendo usage and telemetry data, score them against consistent prioritization criteria, and surface cross-product patterns that no individual product manager would likely catch alone. By the time a product manager reviews an idea, the research is already done.
“We didn't remove humans from the process. We removed everything that wasted their judgment.”
Marcus Alley, VP of Product and Data at Velora
Velora's first experiments weren't perfect. An early Slack agent famously started replying to its own responses, creating endless conversation threads. Rather than treating that as a failure, Alley treated it as a lesson in designing better workflows.
The team introduced confidence scores for every response, built guardrails around when agents could reply, and created monitoring agents that watched other agents for failures. Rather than chasing smarter models, the team focused on building better workflows around trusted product data.
“Pendo built the door. It is a really good door. But a door doesn't walk through itself.”
Marcus Alley, VP of Product and Data at Velora
Velora cut the time to move an idea from initial signal to an evidence-backed roadmap candidate from as long as six weeks to less than a day.
AI agents launched in 30 days
Many of them relying on Pendo's MCP to access live product usage data and customer feedback.
Idea to roadmap candidate
Down from as long as six weeks of manual research and evidence-gathering.
Risk-flagged feedback items actioned
Previously untouched customer feedback surfaced, organized, and turned into actionable product insights.
“I don't log into Pendo anymore. But I use Pendo more now than I ever have.”
Marcus Alley
VP of Product and Data at Velora
Pendo became accessible wherever Velora's employees were already working — through a Slack question, a Notion doc, or the AI agents running in the background. Product intelligence now flows to them automatically. "Every fix is another layer of scope," Alley said. "Not another layer of intelligence."
