A magpie doesn't invent the shiny things it weaves into its nest. It collects them — a bottle cap here, a bit of foil there — and the nest is built entirely from found brilliance. Koe wants you to be that magpie with ideas. Build a place to keep every line that ever made you think "damn, I wish I'd written that," and your content stops coming from a blank page and starts coming from a wall of collected sparks.1
He calls it an idea museum. Marketers call it a swipe file.2 Either way it's the same habit: a single home — a doc, a folder, a note — where you drop ideas the moment they hit, before they evaporate. The counterintuitive claim underneath it: if you don't know what to write, it's not because your mind is dry. It's because you haven't captured enough.3
Strip the museum metaphor and it's a storage-and-retrieval habit with two motions. The first is capture: the instant an idea lands, you jot it down — no judging, no polishing, no "this is too basic."4 The second is sourcing: you don't fill the museum by aimless scrolling, you feed it from a small set of high-idea-density places.5
Koe is specific about the sources. Three to five wells you return to, each so good that visiting them reliably triggers the "I wish I wrote that" reflex.6 His top tier is old or little-known books — where timeless principles live, untouched by trends.7 Second is curated blogs and accounts that do the heavy lifting for you, like Farnam Street, which collects the best ideas for modern intellectuals.8 Third is a handful of heavy-hitting social accounts that always post great ideas — when he's got nothing, he scrolls those for something to react to.9
The museum isn't the goal. It's the inventory that feeds the goal: turning your five to ten best ideas into a thousand posts.10
The deep claim here is unglamorous: writing isn't done from a blank page. It's done from ideas you already stored and now piece together.11 In the Kortex walkthrough, Koe puts it bluntly — if you don't know what to write, it's because you haven't taken down enough notes, haven't captured enough ideas.12
This reframes the whole creative struggle. The blank page isn't a mystical block to push through; it's an empty warehouse. The writer who never runs dry isn't more inspired — they've been depositing into the museum all week. Koe's own setup makes this physical: a capture inbox he texts ideas to himself in, tagged so a "tweet idea" tag filters every raw tweet-seed he's banked.13 When he sits to write, he's not summoning anything. He's shopping his own shelves.
There's a second motion that completes the pipeline. The museum holds raw ideas, but Koe builds a second museum of structures — he breaks down how his best-captured ideas are shaped, then mixes idea and structure like a matrix: take idea one, plug it into idea three's structure, and you've written something new from two things you already owned.14 Capture feeds reuse; the warehouse feeds the assembly line.
The idea museum is the storage layer the rest of Koe's content advice silently depends on. The Big Idea is a withdrawal; the museum is the account it's withdrawn from. One Idea a Thousand Ways is the assembly line; the museum is the parts bin. And Five to Ten Core Ideas Repeated describes what the museum eventually distills into — the handful of refined ideas you spend a career recycling.15
What it hands the vault is a piece of infrastructure thinking: that creative output is downstream of input discipline. That connects it to the AI-collaboration sub-area — specifically The AI Second-Brain Retrieval Shift — and to the practice tradition in Butcher's Dice (Collect, Connect, Emerge), where collecting is the first move of all creative work.
Across the three videos, Koe demonstrates the full pipeline on himself. In Value Creation he describes the database: scrolling Twitter, a tweet sticks out, he pastes it into Notion; reading a book, he highlights, and Readwise syncs the highlight to Notion automatically.16 In Multiple Interests he names the source diet — three to five high-density wells, led by five books he rereads because the ideas are so good.17 In the Kortex walkthrough he shows the retrieval side: a capture inbox, tags that filter all his tweet-seeds, an outline page where stored ideas get assembled into the week's newsletter.18
Read together, the three clips are one organism. Multiple Interests sets the intake (where the good stuff comes from). Value Creation sets the capture (how it gets banked the instant it appears). Kortex sets the retrieval (how it gets pulled back out and assembled). No single video shows the whole loop, but the loop is the actual product — a self-replenishing supply chain for content where the writer's job shrinks to noticing, banking, and recombining. The "almost 1300 document workspace" he flashes on screen isn't a hoard; it's the inventory that makes a blank page impossible.19
It's Tuesday, 9:40pm, you're half-watching a video when a line lands — the speaker says something offhand and your attention snags. That's good. Old you would think "I'll remember that" and lose it by morning. Instead your thumb is already on the capture note you keep one tap away. You type the line, ugly and unedited, maybe four words, and tag it. Done. You go back to the video.
Thursday you open a blank newsletter and feel the familiar dread of nothing-to-say. But you don't sit in it. You open the museum and scroll — Tuesday's caught line is there, sitting next to a book highlight from last month and a tweet you banked in March. You read three of them together and one suddenly rubs against another, and there's your angle. You didn't generate it tonight. You collected it over weeks and only assembled it now. The dread was never about a dry mind. It was a warehouse you'd finally remembered to fill.
Three sources, all the same popular creator across three videos.20 The corroboration is real but not independent — it's one person's consistent method, not three witnesses.
The sharp tension is internal to Koe's own corpus, and it must be flagged rather than smoothed. The idea-museum doctrine says capture everything the moment it appears — if you don't write it down, you lose it, and a dry page means you didn't capture enough.21 But the vault's Trust Ideas Will Come Back (Don't Capture Everything) page — drawn from Koe's own creativity material — argues the opposite: that compulsive capture is a kind of anxious hoarding, that genuinely important ideas resurface on their own, and that the urge to trap every thought can clog the mind it's meant to serve. So Koe holds two contradictory positions across his catalogue: capture relentlessly (the content-machine context) versus trust the ideas to return (the creativity/clarity context). 🚩 He never reconciles them. The most charitable read is that they belong to different modes — capture serves production (you're feeding an output pipeline), while non-capture serves presence (you're protecting a clear, unhurried mind). But Koe doesn't state the boundary, and a reader following both pages literally will be told to write down every idea and to stop writing down every idea. The contradiction stays open.
Koe-the-marketer and Koe-the-creativity-teacher genuinely disagree, and Trust Ideas Will Come Back is where the disagreement is loudest. The idea museum treats a caught idea as a deposit you'd be foolish to let evaporate; the trust-ideas page treats the urge to catch it as a symptom of overstimulation, arguing the worthy ideas come back and the rest were never load-bearing. This isn't a soft tension to gloss — it's the same author prescribing opposite reflexes. Holding both honestly, the resolution that suggests itself is one of purpose: capture is for when you're running a content business and need reliable inventory; trust-it-to-return is for when you're protecting the clarity that lets new ideas form at all. The museum optimizes throughput; the trust principle optimizes the upstream conditions for having ideas worth capturing. A working creator probably oscillates between them, but Koe leaves the oscillation undescribed.
There's a clean convergence with The Big Idea. The two are the same habit named at two ends — the big idea is the catch, the museum is the net. Koe describes them in one breath, and a creator who believes in hunting big ideas without keeping a museum is a hunter with no bag. Where the convergence sharpens into a claim: the ceiling on your big ideas is set by your museum's idea-density, which is set by your three-to-five sources. Better wells, better catch, better output. The pipeline is only as good as its intake.
Plain version: keeping a personal swipe-file and using an AI tool to find old notes look like a low-tech habit and a high-tech feature, but they're solving the identical problem — getting the right past idea back into your hands at the moment you write — and seeing them together reveals what the swipe-file is really for and where AI quietly changes the bargain.
The first handshake is with The AI Second-Brain Retrieval Shift. The retrieval-shift idea holds that AI changes the value of a personal knowledge store from organizing to retrieving — you no longer need a perfect filing system because semantic search surfaces the relevant note on demand. The idea museum is the pre-AI version of exactly this store, and Koe's own Kortex walkthrough shows the seam: he's manually tagging captures so he can filter "tweet ideas," doing by hand the retrieval that AI promises to automate. The structural insight that emerges from holding both: the capture discipline survives the AI shift, but the organization discipline may not. Koe's tags, folders, and Readwise syncs are the manual scaffolding for retrieval — and if retrieval becomes free, the entire middle layer of the museum (the tagging, the careful filing) collapses, leaving only the two endpoints that AI can't do for you: deciding what's worth capturing in the first place, and choosing which retrieved idea to actually use. The museum, read through the retrieval shift, turns out to be three habits stacked — capture, organize, retrieve — and only the first and the judgment at the end are durably human. This also sharpens the museum's real bottleneck: not storage (cheap), not search (becoming free), but taste at the intake door — which three-to-five sources you trust, and which caught lines you bank.
The second handshake is with Butcher's Dice (Collect, Connect, Emerge). Butcher's model makes collecting the explicit first stage of a three-beat creative process: you collect raw material, connect disparate pieces, and let something new emerge from the collision. The idea museum is a purpose-built collect-stage engine — and Koe's "second museum of structures," where he plugs idea one into idea three's shape, is a literal connect move. The insight from putting them side by side: Koe has the collect and connect beats but undertheorizes the emerge beat, while Butcher names emergence as the payoff that collecting exists to serve. Read through Butcher, the idea museum's danger comes into focus — a creator can get so good at collecting and recombining that they only ever produce variations of what they banked, never the genuinely new thing that emerges when collected pieces are allowed to surprise their collector. Koe's "one idea a thousand ways" is recombination; Butcher's "emerge" is transformation. The handshake reveals that the museum optimizes for fluent recombination at some risk to genuine emergence — which loops back, unexpectedly, to the trust-ideas tension: the same compulsive capture that guarantees inventory may crowd out the unhurried mental space where emergence actually happens.
Both handshakes converge on one quiet verdict about the museum. The retrieval-shift reading strips away its middle (organize) and its storage (cheap), and the Butcher reading strips away its recombination (mechanical), and what's left standing in both is the same pair of irreducibly human moves: the taste at the intake door, and the willingness to let collected things surprise you on the way out. Machines can store, search, tag, and recombine; they cannot decide what was worth catching, and they cannot be genuinely surprised by their own collection. So the museum's real value migrates to its two endpoints — curation in, emergence out — and everything in between is scaffolding that AI and habit are steadily automating away.
The Sharpest Implication — The idea museum quietly relocates creative talent from the moment of writing to the months of collecting. If output is downstream of input, then the writer's most consequential skill isn't composition — it's curation: which three-to-five wells they drink from, and whether they bank the spark before it dies. This is liberating and indicting at once. The blank page is no longer a verdict on your imagination; it's a receipt for an empty warehouse. But it also means the work that determines your output happens on Tuesday night with a video paused, not Thursday morning at the keyboard.
Generative Questions