Team norms:
To what goals do we all commit?
• Turn in work on time
• Come prepared to meetings and class
• Have fun!
What are our “ground rules”?
• Assume good intent
• Communicate clearly about availability and time conflicts
• Be respectful and open to ideas others bring up
What platform will we communicate on? How quickly are responses expected?
• Communicate by iMessage
New Questions
• What would you do if you had an extra hour or two at work?
• Where do you face the most amount of friction and why?
• What do you least look forward to in your work?
• What do you most look forward to in your work?
• What is something you would change about your job?
From the reading, I learned to ask about specific situations people found themselves in. That draws out the context around a problem and shows when someone would actually reach for a product.
My list:
• Communication
• Busy Work
• Consensus
AI second pass:
• Confidence in AI-Built Output / Quality Bar
• Deciding What to Build
• Shifting Priorities & Re-planning
• Consensus & Stakeholder Alignment
• Process / Ticket Overhead
• Status Rituals & Business Reviews
• Decisions Under Uncertainty
• Interruptions & Context Switching
• Cross-Team Dependency Planning
• Time to Learn AI / Upskilling
• Role Blurring / Org Design
• Hiring & Talent Pipeline
I think the AI caught more detail than my hand-made list. It pulled out confidence in AI-built output (PM2, PM3) and time to learn AI (PM1, PM2), which my list had no place for. It flattened some things, like merging PM3’s quality-bar concern with PM2’s worry about building the wrong thing, and it counted passing mentions like full stories. It did not invent quotes, but it said business reviews are being solved by AI when PM3 only said that is possible, and it made buckets for problems only one interview raised.
Long List: 6 Problems
Each of these was raised by 2 of 3 PMs.
1. Priorities shift mid-flight and plans have to be rebuilt (2 of 3: PM1, PM2)
“…your entire roadmap could be trashed out. So you probably might have spent weeks discussing and getting people aligned” (PM2)
2. Consensus-building slows and weakens decisions (2 of 3: PM2, PM3)
“…the insistence that there needs to be a consensus to… move forward. And the quality of the decision becomes secondary at that point” (PM3)
3. Knowing whether AI-built output is good enough to ship (2 of 3: PM2, PM3)
“AI is fantastic at making things look and feel… Good enough, until you actually… Test it out” (PM3)
4. Making fast decisions with incomplete information (2 of 3: PM1, PM3)
“…every once in a while, something unexpected happens and we have to rely on judgment, experience, and quick conversations with engineering partners.” (PM1)
5. Low-value process overhead, such as Jira tickets for small approved changes (2 of 3: PM1, PM2)
“Sometimes I have to create a fully detailed Jira ticket explaining the change, the reasoning, the background, and the approvals. At that point, I just want to change the copy.” (PM1)
6. No time to learn AI (2 of 3: PM1, PM2)
“If I have an hour, I pick a card and spend time watching a video, reading a newsletter, or completing a lesson.” (PM1)
TAM: The PM Market
TAM here is annual spend on PM-specific software if every PM worldwide bought it. LinkedIn-based estimates put the number of PMs between about 0.63M employed (Shcheglov, 2023) and 2.1M (Chacon, 2025), so I used 1.0M as the base case. Tools like Jira Product Discovery, Productboard, and Aha! list at roughly $10 to $59 per user per month, so I assumed $40, or $480 per year. That gives a TAM of about $0.5B per year (range $0.15B to $1.5B). Analyst reports on the broader product management software market put it around $8 to $10B (Business Research Insights, Dataintelo), but that likely includes general team tools that aren’t bought only by PMs.
Compared with my cold interview, I found these to be a lot easier. I think I got more insight from the questions and learned to follow up on answers as well. It was also interesting to see different patterns across interviews emerge.
