There is a stat that should terrify every PM: people forget 50 percent of what was discussed in a meeting within 24 hours, and 90 percent within a week. Think about what that means for a product team. The nuance behind a prioritization call, the edge case a customer raised in discovery, the exact reasoning your VP of Engineering gave for pushing back on Q3 scope. Gone. Not because anyone was careless, but because human memory is not built for the volume modern PMs process. This is not a note-taking problem. It is an infrastructure problem.
The pipeline
Rebuild the second brain around conversations, because meetings are the primary knowledge source for product work. The stack is four steps: record, transcribe, embed, query.
Record everything. Stop being selective. Storage is trivial and the cost of missing the one conversation that matters is enormous. Tools like Otter, Fireflies, and Grain join Zoom, Teams, and Meet automatically. Transcribe automatically. Whisper and its derivatives handle accents, jargon, and cross-talk, and most recording tools transcribe under the hood. The output is a timestamped, speaker-identified text file for every conversation.
Embed into a knowledge base. This is where most PMs stop too early. Transcripts are long and keyword search misses context. Vector databases like ChromaDB, Qdrant, or Weaviate convert text into representations of meaning, so you search by concept. Then query like a conversation. Ask what did the enterprise customer in the March pilot say about our onboarding flow, and get the synthesized answer with the actual quotes, not a link to a 40-minute recording.
Why it compounds
Five meetings is a curiosity. Fifty is useful. Five hundred, a year of conversations, becomes something qualitatively different. You can spot patterns no individual would notice: recurring complaints that never quite rise to a formal feature request, gradual shifts in engineering sentiment, the slow divergence between what leadership says it prioritizes and what actually gets resourced. That is intelligence, not information.
Scale it to the team
A personal second brain is valuable. An organizational one is transformative. When every PM runs the same pipeline, a new hire queries the payments-integration history instead of sitting through three weeks of context downloads from six people. When someone leaves, their context does not walk out the door. One useful pattern is the Hemingway Bridge, named after Hemingway's habit of stopping mid-sentence to make it easier to resume: auto-generate a short status summary at the end of each recurring meeting series so continuity does not depend on anyone's memory.
Getting started this week: Day 1, pick a recording tool and enable it for all your meetings. Week 1, review the transcripts and notice how much you had already forgotten. Week 2, set up a simple query layer, even just pasting transcripts into Claude. The gap between I should probably record my meetings and I have a queryable knowledge base of every conversation this year is smaller than you think. Close it.