AI chat brain dump method

I’d like to share an AI-supported workflow that I’ve been using for a handful of months–a lifetime in the world of AI. It’s one that’s proven far more valuable than best practices for individual prompts which have a tendency to change quite a bit over time.

The approach

The approach is simple: I use voice-to-text to brain-dump everything I’m thinking about a topic in one continuous stream of consciousness. Then I ask the AI to organize these unstructured thoughts into a clear, logical format tailored to my audience.

Here’s what a typical prompt looks like:

“I’m going to brain-dump a bunch of unstructured ideas about [topic]. Please help me organize these into [bullets/paragraph structure/outline] that will effectively communicate to [audience].”

Why this works for me

I’ve always struggled with trying to generate ideas AND structure them simultaneously—it’s cognitively exhausting. AI has made this process dramatically more efficient. I’m saving 80-95% of my upfront writing time, and the process feels significantly less draining.

The time investment shifts to editing instead, and I adjust based on context. Quick messages get minimal editing, while things like professional documents get more attention to preserve my voice and ensure quality.

How to implement this

On Windows, I use the Win+H shortcut to activate the built-in speech-to-text in any text field. On my phone, I use the keyboard’s native speech-to-text feature.

I typically dictate directly into Claude, ChatGPT, or Microsoft 365 Copidlot. For ideas that develop more slowly, I’ll capture thoughts in Obsidian over time, then copy everything into an AI chat when I’m ready to structure it.

Keeping it mine

This is where the editing time comes in, and it matters. The AI gives me structure and organization, but the ideas are mine, and I want the final piece to sound like me.

For quick professional messages, I’ll usually accept most of what the AI produces with light touch-ups. For something more substantial like a professional email, document or post, I’m reading every sentence and asking: “Would I say it this way?” I’ll often simplify phrasing, add personal asides, adjust the tone, or restructure sections to match how I naturally write.

I also look for places where the AI might have smoothed over an idea too much or made it more generic. Sometimes my rough, unpolished way of expressing something in the voice dump is actually better than the AI’s cleaned-up version. When that happens, I pull back in my original language. This is also how I keep some of my personal touch and quirky, sometimes eccentric, language along for the ride.

The key is that I’m treating the AI output as a slightly refined and reorganized early draft, not a finished product. It’s done the heavy lifting of organization, but I’m still the writer.

What this isn’t

This isn’t about letting AI do my thinking for me. I’m not asking it to generate ideas about a topic—I’m asking it to organize the ideas I already have.

This also isn’t about optimizing every piece of writing through AI. Some things I still write directly and traditionally, especially when I’m strongly opinionated on the specifics of the text or structure.

And this definitely isn’t a way to avoid the hard work of editing. If anything, I’m spending more focused time on editing than I used to because I’m not exhausted from the initial writing process. The energy I save on drafting gets redirected to refinement.

What this is: a way to streamline idea generation and idea organization, leveraging AI for the part that drains me (tension between specific text and structure during the initial draft) while keeping full ownership of the part that matters most (the thinking and the voice).