Productboard: how Amplitude and Productboard PMs run AI-driven discovery

This article pairs two working practitioners — Frank Lee, Principal PM at Amplitude, and Chris Patton, Principal PM at Productboard — and documents the specific AI workflows each has embedded in their weekly product discovery process. Published in May 2026, it reflects the shift from AI-as-experiment to AI-as-operating-procedure in product teams that have been working with these tools for over a year.

The piece is structured around five workflows that emerged from real practice rather than theory.

Automated weekly product brief. Frank Lee’s team uses agents that scan dashboards and analytics tools each week, surface trends requiring attention, and produce a brief that goes to the team before the Monday sync. The brief replaces a manual data-gathering ritual that previously consumed two to three hours of PM time.

Root cause analysis. When a metric moves unexpectedly, an agent identifies correlations across data sources and generates hypotheses. Lee describes this as compressing the first hour of an investigation from manual SQL queries into a structured starting point — the PM still validates and extends the analysis, but the entry point is sharper.

Session replay analysis. Agents filter behavioral session data to identify clusters of friction and patterns of user intent. This allows PMs to find the most informative sessions without watching hours of recordings.

Opportunity discovery. Rather than relying on periodic roadmap reviews, continuous monitoring surfaces ranked problems with recommended actions. The discovery loop becomes tighter and less dependent on the PM’s ability to be in the right meeting at the right time.

Integrated quantitative and qualitative analysis. Chris Patton’s work at Productboard focuses on Productboard Spark, which combines quantitative metrics with qualitative customer feedback to generate prioritized recommendations. The integration matters because most feedback synthesis tools work on either numbers or text — rarely both in the same workflow.

Both practitioners are careful about the limits of the approach. The article closes with a direct quote from one of them: “Just because we can doesn’t mean we should. Stay grounded in your customer feedback.” The workflows are not autonomous — they surface and synthesize, but a PM judges.

Useful for teams that run continuous discovery, deal with high volumes of customer feedback, or spend significant time on manual data collection before strategic decisions.