Cognitive Sovereignty Self-Audit for Product Managers

1. When you receive user research from Dovetail or similar tools, how do you engage with the raw data?

I read AI summaries and use those summaries to brief stakeholders I read AI summaries but check them against a few interview transcripts before deciding I watch or read at least half the raw interviews myself, then check the AI summary against what I saw I watch or read the majority of raw interviews first, forming my own view before looking at AI output

2. Your prioritisation tool (Jira AI, Notion AI, etc.) ranks features based on inputs you give it. What happens next?

I use the ranking as your final priority list and present it to leadership I check if the top three items feel right, then adjust if they seem obviously wrong I review the scoring logic and adjust weights before accepting the ranking I challenge the ranking by testing it against customer pain points I hold in my head, not just the inputs

3. A customer conversation reveals a behaviour that contradicts your last three research summaries. What do you do?

Note it but trust the AI summaries over a single conversation Flag it to the research team and ask them to check if AI missed something Go back and re-read the original interviews yourself to see what the AI might have filtered out Use the contradiction to rebuild your mental model of the user, then audit all recent summaries

4. When ChatGPT or Claude suggests a product direction based on your roadmap data, how do you validate it?

I share it with the team and move forward if no one objects I think through whether it matches the customer problems I remember I test it against recent customer interviews and my notes from user calls I reconstruct the customer journey from memory and interview data to see if AI missed an edge case

5. How often do you have direct contact with customers outside of formal research sessions?

Rarely or never. I depend on research summaries and reports A few times a quarter in structured calls Monthly customer conversations where you listen more than you guide Weekly or more, and you use these conversations to check if research summaries match reality

6. Your roadmap AI tool flags a feature as low priority. The reasoning shown is brief. What's your next move?

Accept the reasoning and move the feature down Ask the AI to explain its reasoning in more detail Check the raw inputs to see what data points it weighed Cross-check the reasoning against customer feedback and team expertise that may not be in the tool

7. When presenting strategy to your leadership team, how much of your case rests on AI analysis versus your own interpretation?

Most of it comes from AI tool output and summaries Half from AI, half from my own notes and memory Mostly my interpretation, with AI output as supporting evidence Entirely my analysis. I use AI as a starting point but rebuild the argument myself

8. You notice an AI summary of research seems to have flattened out a pattern you remember seeing in the interviews. How often does this prompt you to dig back into the source material?

Rarely. Smoothed summaries are usually fine for decision-making Sometimes, if it feels important to the roadmap Often. You've trained yourself to spot when summaries lose important detail Always. You assume AI summaries remove things that matter and verify before any major decision

Cognitive Sovereignty Self-Audit for Product Managers

Cognitive Sovereignty: How To Think For Yourself When AI Thinks For You