How to Brainstorm Better Ideas With Claude
Most people ask Claude for ideas with an open request, "give me some ideas for X," get back a list of eight reasonable but generic options, and either pick one at random or give up and think of something themselves. The list wasn't bad. It just never got narrowed.
Tip
Brainstorming with Claude is two requests, not one: a wide first pass, then a deliberate narrowing pass against real constraints. Stopping after the first pass is why the results feel generic.
A first list of ideas is raw material, not a decision.
Why the first list is always a little generic
An open-ended request like "give me ideas for X" has no way to know what would actually work for your situation, so Claude fills that gap the same way it fills any other: with a reasonable, broad spread that could apply to nearly anyone asking the same question. That's not a flaw specific to brainstorming. It's the same generic gap-filling that shows up in any prompt missing real constraints, just more visible here because the output is a list you're meant to react to right away.
The fix isn't a cleverer first request. It's treating the first list as a draft round, the same way a first answer to any request is a draft, and following it with constraints that narrow it toward something that actually fits your situation.
Before: ask for ideas once, skim the list, pick whichever one sounds least generic.
After: ask for a wide first pass, then narrow it with the real constraints that first request didn't include.
A process for narrowing a brainstorm
- 1
State the real constraints before the first list, not after
Audience, a hard limit like budget or length, and anything explicitly off-limits. "Ideas for a team offsite" produces something different from "ideas for a team offsite, six people, half a day, no more than $50 per person, nothing outdoors since two people have mobility limits."
- 2
Ask for a wide first pass, more options than you need
Request more than you'll use, ten or twelve rather than three, since a narrow first list gives you less to actually choose between and makes the constraints do less work in the next step.
- 3
Narrow deliberately, by name
Point at specific ideas: which ones are too similar to each other, which ones don't fit a constraint you gave, which two are worth expanding on. This is the step most people skip, and it's where the actual thinking happens.
- 4
Push back on anything still generic
If an idea could have been suggested for almost any version of this request, say so directly and ask for a sharper version, rather than accepting it because it's technically on topic.
I need ideas for a team offsite: six people, half a day, budget of $50 per person, indoors only. Give me twelve options, short descriptions only, no need to fully flesh any of them out yet.
”Once the list is in front of you, the narrowing prompt does the real work, and it should reference the actual list Claude just gave you, not a fresh general request.
Ideas 3 and 7 are basically the same thing with different names, drop one. Idea 9 is outdoors, which doesn't work for us, cut it. Expand ideas 2 and 5 into a few more sentences each, since those are the two I'm actually considering.
”
Inside Claude Tutorial
A wide pass, then a deliberate narrow, applies past brainstorming.
Generating options broadly and then cutting with real criteria is a pattern that shows up in more than idea generation. The app teaches it, and others like it, through short lessons you can apply the same day.
Pushing back on a generic idea, by name
It's easy to let a mediocre idea slide through because it's not wrong, just uninspired. Naming exactly what's generic about it, instead of a vague "give me something better," gets a sharper result than a repeat request that just produces a different flavor of the same broad list.
Common mistake
Asking for "more creative" or "better" ideas without saying what's wrong with the current ones. That's as vague as the original open-ended request, and it tends to produce a list that's different but no more specific.
Idea 4 could apply to literally any team offsite, it doesn't use anything specific about us being a six-person remote team that rarely meets in person. Give me a version that actually reflects that.
”When the brainstorm is for someone else's constraints
Sometimes you're brainstorming against constraints you didn't set yourself, a client's brief, a budget someone else approved, a format a stakeholder already committed to. State those constraints as given facts rather than your own preferences, so Claude treats them as fixed boundaries to work within rather than suggestions worth questioning.
These constraints come directly from the client's brief and aren't up for discussion: budget under $2,000, must launch within three weeks, and no outdoor elements. Brainstorm within those limits rather than suggesting ways around them.
”That distinction matters more than it looks. Claude can't tell the difference between a constraint you're firm on and one you'd reconsider unless you say which is which, and treating a soft preference as a hard limit narrows the first pass more than it needs to be.
Continue reading
- Reviewing Your Own Work With Claude Before Anyone Else Sees It: the same "push back by name, not vaguely" habit, applied to reviewing a finished draft instead of a list of ideas.
- Turning Raw Notes Into a Clean List of Action Items: another everyday task built around asking for a specific, structured output.
- Writing Prompts That Get You a Usable First Draft: the underlying habit, stating real constraints upfront, that this article applies specifically to brainstorming.