Brand Logo
Log in
RU Logo
RU Logo
Home
Courses
Guides
Workshops
Community
Perks
Support
Login

Guide

Use This Two-Step Dictation Strategy To Write Better Docs (Typeless Tutorial)

GuideBeginner

Use This Two-Step Dictation Strategy To Write Better Docs (Typeless Tutorial)

In this guide, you will learn a two-step voice dictation system that immediately makes you write better and improves itself over time.

Required tools

Typeless

Updated

Apr 23, 2026

The Rundown

In this guide, you will learn a two-step voice dictation system that immediately makes you write better and improves itself over time.

Who This Is Useful For

  • Founders and operators who write a lot of memos, updates, and internal docs but do not want them to sound like generic AI
  • Creators and marketers who use AI for first drafts but still need the final version to sound like their actual voice
  • Anyone using Codex, Claude Code, Claude, or ChatGPT who wants a repeatable writing system instead of starting from scratch every time

What You Will Build

image-2180


You will build a repeatable writing workflow where AI creates the first structure, Typeless captures your spoken edits, and an agent rewrites the working draft in your voice.

The key is saving two versions:

  • an untouched initial draft
  • a working draft you can edit freely

That before-and-after trail is what makes the system better over time. Later, an agent can compare the two and pull out the editorial rules you keep applying.

What You Need to Get Started

  • Typeless installed for dictation (free plan works)
  • Codex, Claude Code, Claude, ChatGPT, or another AI tool that can draft and revise documents
  • A document you want to write, like a memo, project update, guide, proposal, or email
  • A simple place to save markdown files, docs, or drafts

Step 1 Set Up Typeless

Download Typeless and go through the setup flow. Give it the permissions it needs so it can use your microphone and paste text into any text box.

image-2178

The important part is the hotkey. Set a trigger you can use without thinking about it. In the recording, we used Option + Space Bar.

Typeless is helpful because you can talk faster than you type, and it cleans up the stutters and rough phrasing better than normal dictation. It also has an AI search hotkey, so you can quickly ask something without switching over to ChatGPT or Gemini.

Pro tip: Use Typeless for comments, not just finished writing. The comments are where your voice comes through most clearly.

Step 2 Have AI Draft the Outline

Open a new Codex or Claude Code session, or use Claude or ChatGPT if that is where you write. Ask for a rough outline or first draft. Do not try to make it perfect yet.

Use a prompt like:

Draft an outline for a short internal memo about [topic]. Keep it concise.

For the demo, we used a made-up project update for an AI software app. The first draft does not need to sound like you. It just needs to give you something to react to.

That is the real move here. AI is good at giving you structure. Your voice comes in during the edit pass.

Step 3 Save the Initial Draft and Create a Working Draft

Before you start editing, tell the agent to preserve the first version.

Use this prompt:

Save this as the initial draft and do not edit it.
Now create a separate working draft that we can revise.

image-2179

This is the secret sauce. The initial draft gives you a clean before version. The working draft is where all the edits happen.

Over time, that difference becomes useful. You can compare the generic AI version against the final version and see exactly what changed: shorter intros, fewer generic phrases, more bullets, different endings, stronger examples, no em dashes.

Step 4 Brain Dump Your Comments

Read the working draft and add a comments section at the bottom. Then use Typeless to talk through everything you would change.

Start with factual fixes first. Then add missing context. Then call out anything that sounds generic, too polished, or not like you.

Your comments can be messy. That is the point.

Example:

The overview needs to be two sentences.
Current progress needs to have bullet points.
Product focus needs to say this version adds social functionality.
Say this version is built for virality.
Make the conclusion less wordy.
Make sure everything is in my voice and verbiage.

This is faster than typing edits one by one. You are basically giving the model a live style guide based on your real reaction to the draft.

Step 5 Rewrite the Working Draft in Your Voice

Once your comments are in the working draft, send the final rewrite prompt.

Rewrite the draft using the comments I left in the draft.
Write it in my tone.
Use my verbiage.
No em dashes.
Preserve the core points, but cut anything that sounds generic.
Make it concise.

The rewrite should look noticeably different from the original AI draft. In the recording, the final version was much less wordy and much closer to the actual direction from the comments.

If it is close but still not right, do another quick Typeless pass. Say the exact issue, then tell the agent:

Make only those edits.

That keeps the model from over-rewriting parts that are already working.

Going Further

The best part of this workflow is that it can improve itself.

Once you have an initial draft and a final working draft, ask an agent to compare them and extract editorial rules.

Use a prompt like:

Look at the initial version and the final working draft.
Pull out the editorial rules for future writing.
Add them to an editorial-rules.md file.
Keep the rules concise and de-duplicate anything repetitive.

You can also turn that into a weekly automation. Have an agent look for new initial drafts and final drafts in the folder, compare them, and update your editorial rules file.

That gives every future writing session more context. Instead of hoping the AI learns your voice over time, you are building your own voice memory in a file you control.

Instructors

Billy Howell

Billy Howell

Educator

Published

April 22, 2026

Categories

General
HomeCoursesWorkshops