AI-Powered "Feedback River" Synthesis
by Shaun Clowes • Chief Product Officer at Confluent at Confluent
Shaun is the CPO at Confluent and previously held CPO roles at MuleSoft and Metromile. He spent six years at Atlassian where he built the industry's first B2B growth team and is a creator of popular Reforge courses on retention and data.
🎙️ Episode Context
Shaun Clowes discusses the stagnation of the product management craft and how to elevate from average to top-tier by focusing outside the building. He explores the nuances of using AI for synthesis versus creation, arguing that data management is the real bottleneck for AI utility. Additionally, he shares his "Bingo Card" approach to career growth and why B2B SaaS moats lie in business workflows rather than UI.
Problem It Solves
Helps PMs synthesize vast amounts of qualitative data without cherry-picking results to confirm their existing biases.
Framework Overview
Instead of using AI to generate ideas, use it to process a continuous stream of customer feedback (interviews, tickets, calls). The key is to prompt the AI to find conflicts with your strategy, not confirmations.
🧠 Framework Structure
Surround yourself with the River: Agg...
Provoke the 'Not': Ask the LLM, 'Wher...
Comparative Strategy Analysis: Feed t...
When to Use
During strategic planning cycles or when analyzing batch customer interview transcripts.
Common Mistakes
Using LLMs to summarize generally without specific adversarial prompts, leading to bland 'hallucinated' agreement.
Real World Example
At Confluent, they use LLMs to ingest inbound requests and semantically group them to see not just keywords, but which underlying concepts are trending up or down in popularity.
Ask ChatGPT to help you find where the customer is probing at the edges of what you're trying to do, where it's wrong, where what you're saying is not what they believe.
— Shaun Clowes