The Bits and Bobs Idea Distillation Loop
Capture everything, filter weekly by resonance, distil, publish — a compounding personal insight engine.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 93%
A weekly cadence that turns raw conversational sparks into published, reusable insight. Capture aggressively in the moment, clean and contextualise within a day or two, filter on Friday by what still resonates, distil at the weekend, publish Monday. The engine is powered by a hard weekly no-meeting block whose only purpose is reflection — and by the discovery that anything you explained the same way to ten people is a document waiting to be written.
Origin
Komoroske's personal system, run for roughly a decade. Built on Compendium (an open-source note tool) which he extended with an embeddings feature to surface similar past cards; distilled output goes into a single ~600-page Google Doc called 'Bits and Bobs'.
Core principles
- 01Capture at the moment the Muse hits — 'I'll write that down later' means it is gone in a month.
- 02Insight comes from reflection on interesting conversations, not from more conversations.
- 03The mundane pointless stuff will take every square inch you give it — you have to defend the space.
- 04Resonance-on-review is the filter, not importance-at-capture.
- 05Allow yourself to be a little illegible in what you publish — you are writing it for you.
- 06Anything explained identically to ten people is a document you have not written yet.
How to run it
- 1
Capture continuously and shamelessly
Write things down mid-meeting, mid-conversation, at 4am. Tell people explicitly that if you are typing it means they said something interesting, not that you are disengaged. Capture even the ones that are only 'maybe onto something'.
Pro tip Speed of capture is a real superpower — Komoroske can pipe an idea straight to his fingertips while still listening, a skill he built transcribing 150 hours of interviews for his undergrad thesis.
Watch out A rough version written in 30 minutes now beats a perfect one you never write. Delay compounds into total loss.
- 2
Process to 'working' within a day or two
Every day or two, go through the raw notes: fix misspellings and add just enough context that the note will still make sense to you in a year. Use similarity/embeddings search to surface related ideas from your past.
Pro tip The 'will this make sense to me in a year?' bar is the exact right amount of processing — no more.
- 3
Protect one meeting-free day for reflection
Block a whole day (Komoroske uses Friday) with zero meetings. Use it to read the documents people actually sent you, and to ask: what thing, had I done it before the week started, would have saved me enormous time and effort this week?
Pro tip That question is the highest-leverage prompt in the system. If you explained the same strategic idea in ten one-on-ones and the framing worked every time, write the 30-minute doc — ten more people will need it and you will not have to be in the room.
Watch out People will give you grief for the meeting-free day. Everyone is busy and always will be — you have to make the time, not find it.
- 4
Filter Friday by resonance
Go through everything captured that week and check only the notes that still resonate — that still seem interesting in some way with a few days' distance. Export the survivors.
Pro tip Resonance days later is a far better signal than excitement at the moment of capture.
- 5
Distil, then publish on a fixed cadence
In a low-pressure window (Komoroske uses weekend nap time), take the surviving notes and distil them into a more durable long-term form. Publish on Monday morning. Let yourself be terse, jumpy and jargon-heavy — you are not writing a clean listicle.
Pro tip The published corpus becomes an asset: he loads Bits and Bobs into LLM project context to think against, and reckons an LLM could draft whole essays from it because the pieces are all in there, just not factored out.
Watch out Do not optimise it for an audience or break it into tweets. The moment it is designed for others it stops being self-reflection and the engine stalls.
In the wild
Roughly a decade of running the loop: 17,248 unpublished working notes in Compendium, feeding a single ~600-page Google Doc that is now so large it lags when opened. The doubling process — capture, clean, Friday-filter, weekend distil, Monday publish — has been running continuously.
→ It became the source of most of his best insights, the origin of the slime mold deck and the kayfabe lens, and a corpus he loads directly into LLM context as thinking fuel.
During Friday reflection he noticed he had explained the same strategic change ten separate times in one-on-ones, with the same framing landing each time. That is the signal: if one framing worked for ten people, ten more people will need it.
→ A 30-minute document, written once, that new people read on their own time without ever needing to involve him — permanently removing him from a recurring cost.
Common mistakes
Waiting for time to reflect
'I wish I had the time' is the standard response, and it is a category error. Everyone is busy and always will be. The mundane will consume every square inch you leave unclaimed — the block has to be defended, not discovered.
Deferring capture
The 'I'll write that down later' impulse loses the idea. A month later you cannot reconstruct what it was or why it mattered. First act of creation happens when the Muse hits, however rough.
Optimising the output for an audience
Restructuring the notes for reach — breaking them into tweets, polishing them for readers — converts a self-reflection engine into a content treadmill. The illegibility is deliberate: readers who bounce off it were not the point.
Is it for you?
Best for
Anyone whose value comes from original thinking — founders, strategists, writers, PMs — who is currently too busy in meetings to have any original thoughts.
Not ideal for
Purely executional roles with no need for novel framings, or anyone unwilling to defend a recurring meeting-free block against organizational pressure.
From the transcript
“I currently have 17248 unpublished working notes”
“what I do on Friday afternoons is I sit down and I go through all the notes I added that week and I just click…”
“the one day I don't take meetings is on Friday and that's the day I read documents carefully that people had sent me or I…”
“if the same idea worked for 10 different people and now in the future there's probably 10 other people that need to hear it”
“the Mundane Pointless [ __ ] will cut take every square inch you give it so you got to make that space to sit back and reflect”
“it's 10 million times if it's like oh I'll write that idea down later you keep on delaying it delaying it now it's a month…”
From the episode
Thinking like a gardener not a builder, organizing teams like slime mold, the adjacent possible, and other unconventional product advice
Alex Komoroske (Stripe, Google)