emily in ai
How to Use AI for Note-Taking
Guides

How to Use AI for Note-Taking

My note-taking system was a disaster for years, half-finished docs everywhere, ideas I could never find again, and then I figured out how to actually use AI to fix it.

I want to tell you what my notes used to look like before I figured this out. There were Google Docs with titles like "ideas - march maybe" and "meeting notes DO NOT DELETE" that I definitely should have deleted. I had voice memos from 2023 that I never transcribed. I had sticky note apps with single-word reminders that meant nothing six months later. It was honestly a little embarrassing.

I'm not embarrassed anymore. And the reason is that I finally figured out how to use AI in a way that actually fixes the specific problems I had, instead of just adding more tools to my already chaotic system.

The Problem With Most Note-Taking Advice

Most productivity advice about AI note-taking is like: "use AI to summarize your notes!" And I'm like, okay, but that assumes your notes are worth summarizing. My notes were not worth summarizing. They were fragments and half-thoughts and things I meant to come back to and never did.

So the first thing I had to fix wasn't my summarization workflow. It was my capture workflow. And that's where AI actually made the biggest difference for me.

Start With Voice, Not Typing

I switched to capturing almost everything by voice first. This is probably the biggest change I made. When I have an idea, in the car, on a walk, mid-shower (okay I wait until I'm out). I record a quick voice memo. The quality doesn't matter. I ramble, I backtrack, I sometimes record the same idea three times from different angles.

Then I drop the audio into a transcription tool. I've been using a Whisper-based one and the accuracy is really good, even with my rambling. What comes out isn't clean prose, it's messy, fragmented, conversational. But that's fine, because that's where the AI step comes in.

I paste the transcript into Claude or ChatGPT with a simple prompt: "This is a voice memo I recorded while thinking through an idea. Pull out the key points and organize them, but keep my original phrasing where possible." And it does exactly that. I get back something that actually sounds like me, organized in a way I can use.

This has become my primary brainstorming method. It's faster than typing and it gets me to clearer thinking because speaking forces you to actually finish sentences, which typing doesn't always do.

Meeting Notes: The Thing That Changed Everything

Okay here's the workflow I'm most proud of. For calls and meetings, I use a transcription tool that runs in the background and captures the full conversation. Afterward I feed that transcript to an AI with a very specific prompt:

"Here's a transcript from a meeting. Please give me: (1) a two-sentence summary of what we discussed, (2) a bulleted list of decisions we made, (3) a bulleted list of action items with who owns each one, and (4) any open questions that came up but weren't resolved."

That's it. Takes about ten seconds. And what I get back is better than any meeting notes I ever took manually, because I was always too busy trying to keep up with the conversation to capture it accurately.

I'm not going to lie, the first time this worked really well I just stared at the output for a second. It felt like cheating. It's not cheating. It's just actually useful.

How I Organize Everything After

I keep all my notes in Notion. This is personal preference and I know some people hate Notion, which is fine, this workflow works in Obsidian or Roam or wherever you already live.

The thing that makes this work in Notion specifically is the AI search. I don't have to perfectly tag or organize every note anymore, because I can just ask Notion AI "what did I decide about X" and it finds it. This has genuinely removed the biggest barrier I had to taking notes in the first place: the fear that I'd never find the note again anyway, so why bother?

That said, I do have one organizational habit I kept: everything gets a date and a one-line context note at the top. Something like "Jan 25, thinking through Q1 priorities after call with Sarah." That two-second habit makes the AI search way more accurate because it gives the AI more context to work with.

The Weekly Review Step

Every Sunday I do something I used to think sounded really boring and now I genuinely look forward to. I collect all the notes I took that week, meeting notes, voice memo transcripts, random ideas, paste them into a single doc, and ask the AI: "What were the recurring themes in my thinking this week? What decisions did I keep coming back to? What seems most important based on what I was spending mental energy on?"

The answers are often surprising in the best way. I'll see a theme I didn't consciously notice, like I was worrying about the same thing in three different contexts, and that awareness changes what I prioritize the next week. It's become one of my favorite parts of my workflow, and it takes maybe fifteen minutes.

What Doesn't Work

I tried using AI to take notes in real-time during conversations, like having it listen and summarize live. This was distracting and the summaries were worse than doing it after the fact. Real-time AI note-taking might get better, but right now I think it's a gimmick.

I also tried using AI to pre-organize my notes before I even had them, like, setting up templates for every type of note and having AI fill them in. Too rigid. My thinking doesn't fit into predetermined templates and trying to force it there made me stop taking notes at all.

The Simple Version

If you want to start small: just try the meeting notes thing. Record your next meeting, transcribe it, and ask an AI for the summary and action items. Do that three times and I promise you'll be a convert. Start there and figure out the rest as you go. That's basically what I did.

Emily in AI

Emily in AI is a plain-English guide to AI tools, tips, and beginner guides. Every tool gets tested and written up without the hype or the jargon, so you can figure out what actually helps. New posts every week.

About Emily in AI →