Context
In the tech world, many meetings now have an AI note-taker, and some people even wear devices that record every conversation. Petra Wille and Teresa Torres, two product coaches and hosts of the show, explore what that does to working relationships. Petra had a colleague pull up a transcript to correct how she remembered her own past opinion. Teresa uses Granola transcripts on her one-on-ones to check her own memory. It matters to PMs because so much of product work is built on conversations, agreements and shared memory of what was said.
The Big Idea
A permanent record of conversations is most useful as a mirror for your own memory and a tool for shared learning, and most dangerous when it is used to hold people to opinions they have since outgrown.
Both hosts found that transcripts exposed how unreliable their own memories are. That value only holds if teams first agree that opinions are allowed to change and that the records themselves can be wrong.
Key Insights
Transcripts expose your own faulty memory
- What: Teresa often feels she already told someone something and gets frustrated that they have not acted on it. She now forces herself to check the Granola transcript of that one-on-one.
- Why it matters: "More often than not, we didn't talk about it before." The frustration came from her, not from the other person.
- Example: she was sure her husband owned two copies of a band T-shirt, because she could picture the one she remembered. He owned one, and the shirt in her head did not exist.
Thinking something feels like saying it
Teresa's theory: our brains are not good at telling apart what happened inside our heads from what we said out loud. When we think we told someone something, it may be because we planned to tell them. For a PM this is a reason to assume the team can only see what was actually made visible, not what you were thinking.
Being checked by others feels different
- Using a transcript on yourself is a private, safe experiment.
- Petra had it done to her. A colleague said he had the meeting transcribed and that her opinion a few weeks earlier was slightly different from what she remembered.
- Her brain was sure of its version, and for a split second it felt weird. The gap was small: the arguments still held, but the nuances differed.
- They stayed friends. Petra says she had not asked for the check, and that made the difference.
Should past opinions bind us?
Teresa raised the bigger problem with recorded conversations: should we hold people accountable to what they said earlier, and do we give them permission to evolve? Her view is that opinions should be evolving at all times. She points to readers who note that two of her articles, written seven years apart, disagree, and her answer is that she hopes you keep learning as well. Petra adds that small shifts in an argument can be a human strength, because she may have learned something new or read another article in the meantime.
Ask why the opinion changed
Petra's colleague asked her why her view had moved "two millimeters". That turned an awkward moment into a useful one: she had to explain her reasoning, and it showed her how her opinion emerged. The record was a starting point for a conversation, not a gotcha. The lesson is that what matters is what someone thinks now, and the interesting part is the reason for the change.
Records can be wrong
- First layer: the transcript is itself a translation of speech, and may be inaccurate.
- Second layer: an AI then pulls information back out of its memory and interprets it again.
- Related point: both hosts have seen AI quote them wrongly, even though it had access to their writing.
- Practical limit: the person checking may be reading Granola notes and not the actual transcript, and few people will make the effort to verify.
Mental Models & Frameworks
Records as a shared learning tool
Teresa suggests that, used well, recorded conversations are a collaboration tool and not only a personal one. It works if two preconditions hold: the team accepts that people can change their minds, and the team accepts that records may not be 100% accurate. Then it becomes safe to say "didn't we talk about this last week?" and follow a simple sequence:
- Look at the record together.
- Ask what our opinions were back then.
- Ask what we have learned since.
- Say where we stand today and why it changed.
The aim is a shared experience of what changed the team's perspective.
Adult development stages (theory of mind)
- Petra mentioned stages of adult personality development, which she wrote about in her email course. She named the researcher as Jane Loevinger (the transcript spells it "Levinger") and tied the roots to Chris Argyris and his action science.
- Teresa links these models to theory of mind: a child realizing that their experience is not the same as someone else's. The adult models look at how that ability grows, and people vary a lot in how well they take other perspectives.
- Petra reads the stages as nested: you cannot reach a later one without the learnings of the earlier ones. Later stages come with awareness of your own memory flaws and acceptance that two people can hold different truths.
- Teresa is not convinced the stages are real, since they are still theoretical models, and she dislikes the idea that later stages are "better". She prefers a toolbox picture, where earlier tools can fit some situations better than later ones. Petra disagrees and ties it to life experience and neurological flexibility. They left it open and floated a future episode.
Trade-offs & Nuance
Useful mirror, possible weapon
Petra notes the same record that helps a team can be weaponized: "you said this on this date, why are you changing your mind?" Her suggestion is to use transcripts on yourself first and see how it feels before using them on others or on a team. The trade-off is between honest shared memory and a culture where people feel watched and policed.
Recording is a tech bubble habit
Petra points out that the "everything is recorded" world exists mainly inside the tech bubble. Both hosts say wearable recorders, like Meta glasses, feel creepy to them, and Petra hopes they get banned in Europe at some point. The nuance is that how welcome recording is depends heavily on whether people agreed to it.
Common Mistakes
Mistake: Using transcripts as a gotcha
- Why it happens: it is easy to search a transcript and find a quote that contradicts what someone says today.
- Why it hurts: it assumes people's views should not move, which is both unrealistic and bad for learning.
- Better approach: ask what changed, so the group learns how the opinion evolved.
Mistake: Trusting the notes blindly
Petra suspects the colleague who corrected her was looking at AI-generated meeting notes, not the full transcript. Teresa agrees we should not always assume notes are correct. Treat a summary as a lead to check, not as proof.
Practical Application
Check your frustration against the transcript
Next time you think "I swear we already talked about this," open the transcript of that one-on-one before you say anything. Teresa reports that more often than not the topic had not come up, which reframes the problem from the other person to your own communication.
Test it on yourself first
Before you quote a transcript back at a teammate, read your own old transcripts and see how often your memory differs from what you actually said. Petra's advice is to feel what it is like on your side before using it on others.
Run a "what changed" conversation
- Do: when a record shows an opinion has shifted, ask the person what they learned in between.
- Then: agree on where the team stands today and why.
- Why it works: it gives the team a shared story of what changed, and turns a possible accusation into a learning moment.
Say the records may be wrong
When you pull up a transcript or summary in a meeting, say out loud that it may be inaccurate. Teresa names this as a precondition for it feeling safe to say "can we look at the record?"
Questions to Consider
- When did you last feel sure you had told a colleague something, and did you check whether you actually said it?
- Does your team feel free to change its mind on a decision made weeks ago, or does an old meeting transcript make people defend the earlier view?
- If a teammate corrected your memory of your own past opinion using an AI note-taker, how would you want them to raise it?
- Who in your meetings has not agreed to be recorded, and how might that change what they say?
Bottom Line
When almost every conversation is transcribed, your memory is no longer the record, and that can teach you how often it is wrong. Use transcripts on yourself first, grant others the right to evolve their opinions, and treat the record as the start of a conversation about what changed, never as a weapon.
Resources Mentioned
| Resource | Type | Why it was mentioned |
|---|---|---|
| Jane Loevinger's stages of ego development (Wikipedia) | Research model | Petra's example of adult development stages, linked in the show notes for further reading. |
| Chris Argyris's action science | Research field | Named as the root that adult development research grew out of. |
Tools & Products
| Tool / Product | What it does | Why it was mentioned |
|---|---|---|
| Granola | AI meeting note-taker and transcriber | Teresa uses it on all her one-on-ones to check her memory against what was said. |
| Meta glasses | Wearable device that can record conversations | Both hosts say they find constant recording by wearables creepy. |
People to Follow
Teresa Torres
Co-host of the show. In this episode she shares how checking her Granola transcripts changed the way she handles frustration with colleagues.
Petra Wille
Co-host of the show. She shares the experience of having a colleague correct her recollection of her own opinion using a transcript.
Notable Quotes
"More often than not, we didn't talk about it before." (Teresa Torres)
"We need to grant each other the right of evolved opinions." (Petra Wille)
"Why does it matter what you might have said three weeks ago? Doesn't it matter more what you think now?" (Teresa Torres)
