Ideas do not arrive when you sit down to write. They arrive in the shower, on a walk, halfway through a conversation about something else — and then, later, when you are at the desk with an hour and a blank page, nothing.
That is not a creativity problem. It is a scheduling mismatch: the arrival and the use happen at different times, and if you do not bridge the gap, every session starts from zero while your best material evaporates in the queue at the supermarket.
What actually needs catching
Not "ideas", which is too vague to act on. Four specific things, and they behave differently.
The angle. Not a topic — a take. "Habit trackers are the worst-suited category to being their own app" is an angle. "Habit trackers" is a topic, and topics are worthless because you already have infinite topics.
The first line. Sometimes an opening arrives fully formed. Write it exactly as it came, because you will not reproduce the rhythm later, and a good first line is often the whole piece in compressed form.
The observation. The specific thing you noticed. A detail, an overheard sentence, an odd fact, a moment. These are the raw material that makes writing concrete rather than general, and they are the fastest to fade because they were never verbal to begin with.
The connection. "This is the same argument as that." Connections between things you already know are where most original writing comes from, and they arrive unannounced and leave immediately.
Notice what is not on the list: research, links, quotes from other people. Those are collectable at any time, which is why people collect them — it feels like preparation and it is the part that is never the bottleneck.
Capture in the form it arrived
The most common mistake is tidying at capture time.
Someone has a rough, alive, slightly incoherent thought and writes down a neat summary of it. The summary is accurate and dead. Six months later it reads as a topic, and the thing that made it interesting — the specific irritation, the odd angle, the exact phrasing — is gone.
Write it messy. Write the sentence as it occurred to you, including the swearing, the incomplete clause, the "something about how...". The mess is the information. It is what will let you recognise, later, why you cared.
The corollary: capture has to be fast enough to allow mess. Anything that requires choosing a category, a tag or a location will produce tidied output, because you will compose while you file. The four-second window applies here more than anywhere.
One bank, not a filing system
The temptation is to organise ideas by theme, project or type. Resist it, for a specific reason.
Ideas are not useful in the category you assigned them. They are useful when they collide with something else, which is a thing that only happens if they are in one place where you read past the ones you were looking for. A well-organised idea file returns exactly what you searched for and nothing else, which is the opposite of what you want from it.
So: one place. Either one note per idea in a flat collection, or one long note per month that you append to. Both work. The one that does not work is eleven notes by theme, because assigning a theme is a decision at capture time.
Add a date and, if you can, one word about where you were or what prompted it. That context is often what makes the idea legible again later.
Reading it back is the practice
Capturing is easy. Rereading is the part that makes the bank worth having, and almost nobody does it.
Once a month, read the whole thing. Not to find something — to reload it. Ideas that have been read recently are the ones that surface while you are writing about something else. An unread bank is a graveyard, however good the contents.
Expect most of it to be bad. Genuinely. Two thirds of anything captured in the moment turns out to be nothing, and that ratio is fine and normal. The bank works on volume; the price of catching the good one is holding the mediocre ones.
Notice repeats. The thing you have written down four times in different words is trying to tell you something. Repetition across months is the single strongest signal a bank produces, and it is invisible unless you reread.
Delete freely. The same eviction discipline that keeps a someday list readable. If it has been there two years and produces nothing on reading, it goes.
The gap between an idea and a draft
The step nobody talks about, and it is where most banks stall.
An idea is one line. A draft needs a shape. In between there is a small, specific piece of work: taking the line and writing three or four sentences about why it matters and what the argument actually is.
Do that when you first reread it, not when you sit down to write. It takes five minutes, it is much easier than starting a draft, and it converts an idea into something you can pick up on a low-energy day. A bank of one-liners requires inspiration to use. A bank of four-sentence sketches does not, and the difference in how often you actually write is substantial.
The honest version
A large idea bank is not evidence of anything. Plenty of people have four hundred captured ideas and no finished work, and the capturing became the hobby — which is the same failure as building the system instead of using it, applied to writing.
The bank is worth exactly as much as the rereading. If you are not going to reread it monthly, capture less and write more.
Cyanote suits the messy-and-flat approach: ⌘N for a new note, ⇧⌘D for today's daily note if you would rather append than create, and no requirement to file anything anywhere. ⇧⌘F searches the body of every note you have ever written, [[ links an idea to the piece it eventually became, and notes nest into sub-pages when a sketch grows into something. It is one local database on your own Mac — which for half-formed thoughts you would not want anyone to read is a reasonable place for them to sit.