Good Talk: You Say Potato, I Say My Last Name Wrong
December 9, 2025Good Talk: Remain Memorable When AI Takes the Minutes
By: Michael Chad Hoeppner
Published: February 4th, 2026
You are being notated.
You are being collated.
You are being condensed.
And synthesized.
And summarized.
And (worst) deleted.
As companies are relying on AI tools more and more to do the mental lift of crystalizing the essential epiphanies, next steps, and outcomes of meetings, LLMs are reviewing everything you say, deciding what’s important, and then either amplifying or relegating your contributions.* So if you want your recommendation for business strategy (or your suggestion for where to host the company barbecue) to be taken seriously, it’s time to grapple honestly with what AI recording means for how you show up at meetings.**
Let’s first consider how these listening robots digest differently than we do. They don’t…
- have recency bias. It doesn’t matter if you say a brilliant opener, a riveting closer, or your genius insight is buried in the murky middle of the meeting. LLMs don’t overindex on timing. They hear and digest it all. So it will no longer work to be the underprepared kid in the college lecture trying to rig the grading system by reading the first five pages of the homework and making a killer point in the first moments of class, never to be heard from again.
- care who’s speaking. Status, hierarchy, room position—so long as you’re not so far away that the software can’t hear you, and/or the AI has been trained to factor in identity and role of the speakers—none of that affects the ingesting and organizing the robots do.
- listen via chronology, priority, or immediacy. They listen entirely. Imagine that difference: we forget something people said in moments; they never forget. Never.
Content
What not to do:
Here is the horrible advice I will not give you: “Prepare More!” As tempting as it may be to listen to the voice of that brutal taskmaster inside your brain and that relentless chorus, let me offer some liberation: ignore that voice. Not because it’s not a valid suggestion, but rather because you’re probably not going to do it. And, it’s the same thing you already know.
Instead I will give you this advice: train yourself to be more disciplined to say smart stuff in real time.
How to train:
Use the embodied cognition drills GK Training teaches to help people around the world. The best two for the above goal are the Lego drill (you can find that drill here; or demoed by a famous actress; or read about it as described in this fictional thriller that featured a character based on me; or Chapter 6 of my book), and Finger Walking (demo vid here; or read in Chapter 7). These are so useful in the age of AI that I’m more than happy to share them; if you don’t have time, money, or inclination to get my book, just email me and I’ll send you free PDFs of those chapters!
Delivery
Enunciation
Wait; AI doesn’t care about delivery, right?
It’s tempting to think of AI like a notetaker at a deposition: it faithfully captures every word, but doesn’t care about the sound of those words. Superficially, this is true. But the conclusion is the wrong one. Here’s just one of the reasons why: If you use your delivery skills—posture, breath, enunciation, the 5 P’s of vocal variety, etc.—to underscore the importance of a point, your idea will land with more impact on the other humans in the room and likely merit more discussion about your point. So your idea will occupy more real estate in the meeting, and that larger footprint will probably be evidenced in the AI synthesis later. Capturing your audience’s attention still matters. Don’t Dismiss Delivery!
AI notetakers have many more implications than the few I’ve outlined above, but different than an LLM, you do have limited powers of attention and stamina, so I will refrain from overwhelming your circuitry.
The last thought I will leave you with is this: remember—you are not a robot. So be a human: breathe, take time to think, empathize with and look at your colleagues/friends/co-travelers, and take not yourself so seriously (the Yoda-like grammatical construction there is on purpose; Thought Suppression is a thing).
To err is human. So don’t make the err-or of thinking you need to be flawless.
With humanity,
Michael Hoeppner and the GK Training Team
P.S. AI did not write this piece for me!
* It is very much an open question what companies will do with these transcripts, syntheses, and reports. It’s easy to imagine a world in which it all amounts to more meeting summaries that don’t get read or acted upon. But let’s give their applicability the benefit of the doubt given their ever-increasing ubiquity.
** One could cheat the system; one could just repeat oneself over and over and/or dominate the meeting with the sheer number of words spoken (translation: TALK A LOT) and this repetition and volume would likely show up in the AI synthesis of the meeting. You would get a lot of exposure and representation in those notes. But one only has to imagine the impression that behavior would make on your colleagues to see it’s a counterproductive strategy. That doesn’t mean some may not adopt it. Consider yourself warned.