There's a comforting story we tell about note-taking, which is that if you just paid closer attention and wrote a little faster, you'd capture the meeting. It isn't true, and it's worth understanding why before you blame yourself one more time for the thing you forgot to write down. Notes don't fail because you lack discipline. They fail because writing notes is a sampling process — you're recording a few moments out of a continuous conversation — and the sample you take is biased in ways you don't choose and can't see.
Around the fire, nobody scribbled while the story was being told. They listened, and the parts that mattered were carried by the group, not by one person's hand racing to keep up. Hand notes ask you to do the impossible: listen fully and transcribe selectively at the same time. The selection is where it all goes wrong.
A note is a sample, and you don't control the sampling.
Think about what physically happens when you take notes. A conversation is arriving in real time. You can't write as fast as people talk, so you write some of it — and the part you write is whatever your attention happened to land on in the second you reached for the pen. That's a sample. And a sample is only as good as the process that draws it.
The trouble is the process drawing your sample is your attention, and your attention has a strong, predictable bias. It captures what's vivid over what's important. It captures the thing said with emphasis over the thing said in passing — even when the thing said in passing was the actual commitment. It captures the start of the meeting, when you're fresh, far better than the end, when the real decisions tend to get made. You are not sampling the conversation evenly. You're sampling it through a filter you didn't install and can't turn off.
The four commitments your notes reliably miss.
Once you see note-taking as biased sampling, you can predict exactly which information falls through. It's not random. It's the same four kinds, every time.
The late commitment. The most consequential thing in a meeting is often the last thing — the “okay, so I'll have it to you by Thursday” said while everyone's standing up and closing laptops. Your notebook is already shut. Your attention has already left. The decision the whole meeting was building toward is the one your sample is least likely to contain.
The casual commitment. “Yeah, I'll loop in legal.” Tossed off, no emphasis, gone in three seconds. It didn't feel like a commitment in the moment, so your hand didn't move. Three weeks later it's the thing nobody did, and nobody can prove was promised.
The conditional commitment. “I'll send the revised numbers once the client confirms scope.” You write down “send revised numbers.” The condition — the entire reason it's still sitting on someone's desk — doesn't survive, because conditions are quiet and your pen is slow.
The inconvenient commitment. This is the uncomfortable one. We are quietly less likely to write down the thing we agreed to do, especially the thing we'd rather wriggle out of later. Not dishonestly — just human. The sample bends, gently, toward the version of the meeting that's easiest to live with.
None of these are exotic. They're the substance of how work actually gets agreed. And they're systematically underrepresented in every set of hand notes ever taken, because the sampling process is biased against exactly the moments that are quiet, late, conditional, or inconvenient.
Why “take better notes” can't fix a sampling problem.
You can't out-discipline a biased sample. If the mechanism is “write down what catches your attention, in the order it catches it,” then writing more just gives you more of the same bias at higher volume. You'll capture more of the vivid, emphasized, early material — and still miss the quiet thing said at minute 38. The fix isn't a better sampler. It's not sampling at all.
A recording doesn't sample. It takes the whole conversation, evenly, start to finish, the casual aside weighted the same as the headline decision. Nothing is filtered out at capture time because nothing is being selected at capture time. That's the structural difference, and it's the only thing that actually closes the gap. For the deeper version of why even the summaries built on top of recordings can re-introduce a bias of their own, the wrong-shape argument is its own piece.
What Bonfiyah does instead of asking you to sample.
When you record a conversation with Bonfiyah, the capture is complete and unbiased by construction — it's the whole thing, transcribed and speaker-separated. Real-time transcription runs on-device; audio leaves your iPhone only for the optional cloud-transcription pass you control. The selection that matters then happens after, against the full record, where it can be done deliberately instead of in a panic with a pen.
Promise Tracker reads the entire conversation and pulls out every commitment — yours and theirs — including the late one, the casual one, the conditional one, and the inconvenient one your own attention would have quietly skipped. Each is attributed to who said it, dated if there was a date, with the exact quote a tap away. The condition comes along with it, because the transcript kept the whole sentence, not the fragment you'd have had time to scrawl.
Action Items does the parallel job for tasks — the to-dos that surfaced in the conversation, pulled out and listed without you writing a single one down by hand. Between them, the work the meeting actually produced is captured from the full record, not from the lossy sample your hand could manage in real time.
What this isn't.
This isn't an argument that you should stop thinking in meetings, or that the act of writing something down has no value. Sometimes the point of a note is the thinking, not the record — the jotted question, the arrow connecting two ideas, the thing you scribble to stay engaged. Keep doing that; it's good for you. The argument is narrower and, I think, harder to dodge: if you're relying on hand notes to be the durable, complete, trustworthy record of what was agreed, you're relying on a biased sample to do a job it structurally cannot do. It will keep missing the late, quiet, conditional, inconvenient things, and it will keep feeling like your fault when it does.
It isn't your fault. It's the method.
See what your notes have been missing.
The honest way to feel this is to run them side by side. Record your next meeting with Bonfiyah, take your usual notes the way you always do, and afterward compare your page against what Promise Tracker pulled from the full conversation. Count the commitments on each. The gap between them is the sample you've been losing this whole time — and it tends to be exactly the commitments that come back to bite.
Bonfiyah is free to start. A good record isn't the part you managed to catch. It's the whole conversation, kept — so the quiet thing at minute 38 is still there when it turns out to be the thing that mattered.
— Richard
