Skip to main content
Using Claude

Brainstorming Well with Claude: Separate Diverge and Converge

Brainstorm with Claude using a two-step diverge-converge flow: generate many first, then narrow by criteria.

By
🌐 This article was machine-translated and may contain inaccuracies. Read the Korean original if in doubt.

Brainstorming is one of the easiest ways to put Claude to work. People tend to stall because they judge ideas the moment they think of them, but you can ask an AI to "hold judgment and just generate a lot." The key is separating diverging from converging.

🟢 Model references match the current lineup · model notice · Fable subscription
🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.

Fable 5 and 5.1 subscription (updated September 7, 2026): Claude Fable 5.1, released September 1, 2026, is the current Fable model and Fable 5 is now legacy. Plan terms are the same for both — Max and Team Premium plans include Fable at up to 50% of the weekly usage limit; Pro and Team Standard use usage credits (

🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.
0/M input, $50/M output tokens). The one-time
🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.
00 credit applied only to the Fable 5 transition and is not offered for 5.1. Some coding and debugging requests may be answered by an Opus model due to a security classifier (both models). See the Fable 5.1 guide and the Fable 5 availability guide for details.

Separate diverge and converge 1. Diverge — generate many 2. Converge — narrow by criteria 1 2 3

Why AI brainstorming works

People generate fewer ideas because they immediately think "that will not work." You can tell an AI to stop evaluating and quickly produce many candidates, and it is easy to try again from different angles. It will even offer odd combinations you would not reach on your own, which makes it especially strong at the diverge stage.

Stage 1 — Diverge: generate many

First, increase the volume. Do not weigh good against bad; gather as many candidates as you can.

  • State a quantity — ask for a number, like "20 different approaches." A number pushes the ideas further out.
  • Provide conditions — target, budget, tone; context yields more usable candidates.
  • Hold judgment — say "do not evaluate at this stage, just list them."

Stage 2 — Converge: narrow down

Once you have enough, set criteria and cut down.

  • Give criteria — "top 3 by novelty and feasibility."
  • Ask for grouping — clustering similar ideas reveals the bigger picture.
  • Develop — turn the chosen ideas into concrete plans.

Shift the perspective and retry

The same topic yields new ideas when you change the angle. When you get stuck, reframe the request like this.

Reframe and retry Opposite weak points surface Constraints "budget one-tenth" Time shift "from a 10-year view" Audience "a different user"

Twisting the conditions surfaces options the original framing hid. You can also take an idea from one frame and run it through another.

Caveats

AI ideas are a starting point, not a conclusion. Facts, feasibility, and legal or ethical concerns must be verified separately by a person. A plausible-looking suggestion may have no basis, so confirm before acting on important decisions.

Just separating diverge from converge changes results dramatically. For more, see the official Anthropic documentation.

Was this helpful?

Keep reading

Claude's Skills Explore Tab Explained — What 60 Plugins Do, How to Use Them, and Whether They Run Automatically (Sept 2026)

What each plugin listed under Settings → Skills → Explore does, how you call it, and whether it can run with nobody around — based on the SKILL.md files of 60 installed plugins and Anthropic's official docs.

65

Can Claude Really 'One-Click' Write a Web Novel? What's Realistic

There's a hope that feeding Claude a premise is enough to get a finished serialized novel, but the tasks Claude is actually good at and what still needs a human writer are different things. Here's a realistic workflow using real features like Projects and Styles.

38

When Your Claude Character Drifts Mid-Conversation: Keeping Persona Context Stable with Projects

Long conversations can wear down a carefully built character setup. Here's how Claude's project instructions and knowledge features keep a persona consistent over time.

37

How to Save Claude Usage: A Practical Guide for Research and Experimentation

A few small changes to how you chat can meaningfully cut your Claude usage. Here's how project caching, turning off extended thinking, Claude Code commands, and the September 14 weekly limit change all fit together.

89