Resonant + Medium
You already said it.
Now write it down.
The Medium posts worth writing are usually things you've already explained out loud — to a teammate, to a friend over coffee, in a DM you sent at midnight. The verbal version already exists. Voice is how you stop losing it.
Resonant transcribes on your Mac. Press a key inside the Medium editor and speak the post the same way you'd explain it to a friend. Clean prose appears in the draft. No cloud audio, no extension.
Used by professionals at
Typed vs. dictated
The written version is careful.
The spoken version is honest.
Postmortem for a public write-up
“On Tuesday our payment processor had an outage and we lost about two hours of transactions. Here's what we learned.”
“The thing I didn't want to admit on Tuesday morning is that we'd been warned. Two weeks earlier, our payment processor had quietly shipped a rate-limit change in their staging environment, and the release notes for it were buried in a changelog I was supposed to read on Fridays. I don't read that changelog on Fridays. So at 9:47 on Tuesday morning, our webhook ingestion queue started throwing 429s, and by 9:52 we'd lost visibility into about forty minutes of real transactions. What follows is the timeline, the detection gap that made it worse, and the three things we're changing permanently so that the next vendor changelog doesn't take us out again.”
Tutorial introduction
“This tutorial shows you how to set up a CI pipeline for a TypeScript monorepo using GitHub Actions.”
“Every TypeScript monorepo tutorial I've ever read assumes you already understand why the standard GitHub Actions setup won't work for you. That's the part I want to fix. In this walkthrough I'm going to start with the specific failure mode that made me learn this in the first place — a pull request that passed CI, merged cleanly, and then broke production because the changed package wasn't actually being tested. We'll look at why that happens, how turborepo and nx handle it differently, and then we'll build a working pipeline from scratch. By the end you'll have CI that only tests what actually changed, and you'll understand why every line of the config is there.”
Hot-take essay
“I think most engineering managers should still code. Here's why.”
“The standard advice is that once you become an engineering manager, you should stop coding, because coding is a contributor's job and you're not a contributor anymore. I've managed engineering teams for seven years and I think this advice is mostly wrong — not because managers should be shipping features, but because managers who stop writing any code at all lose the ability to distinguish hard problems from problems that only sound hard. That distinction is the entire basis of good engineering management, and you cannot maintain it from a distance. In this post I want to argue for a specific, narrow form of managerial coding — not sprint work, not production code, but something else — and explain what I've seen happen when managers ignore it.”
The insight
The best version of the post
already exists as talking.
Medium is where practitioners share what they've actually learned. Incident writeups, framework comparisons, hard-won opinions about hiring or systems or design. The things worth reading are almost always things the writer has already explained verbally — in a Slack thread, on a call with a skeptic, in a DM where a friend pushed back and forced the argument into its sharpest form.
That verbal version tends to be better than the written version you'll sit down to draft later. It's concrete. It has the specific example. It hasn't been sanded down by the self-conscious voice that shows up the moment you open an editor and start typing for an audience. And it usually evaporates within a week, never to be recalled in quite the same words.
Voice bridges the two states. You can dictate the post the way you'd explain it to someone who asked — and you get text that carries the explanation-to-a-friend tone into the Medium editor, where you can polish it without losing the version that was actually alive.
The moment you switch from explaining something to a person to writing it for a public audience, your voice changes. Sentences get longer. Hedges appear. The concrete example gets replaced by a general claim. This is the Medium tax.
Most posts pay it. The best ones don't.
Speaking bypasses the internal editor that formalizes everything. You end up with prose that reads like you're explaining the idea to the reader, not performing it — which is, not coincidentally, how every great Medium post sounds.
The post ships with its original energy intact.
Where it fits
Six kinds of Medium posts
that start as talking.
Things you've already said in Slack
The technical explanation you gave a teammate last week is probably the cleanest version of the post you're about to write. Voice lets you retell it into Medium while the phrasing is still fresh.
The DM that turned into a rant
Every writer has that moment where a reply to a friend grows into 600 words of actual argument. Next time, route the rant into the Medium editor instead of losing it to a DM thread.
Postmortems and incident stories
The honest version of a postmortem is the one you tell in the bar afterward. Dictating the draft preserves the honesty that a careful keyboard-first write-up usually sands off.
Conference talk adaptations
You already gave the talk. Now you just have to write the Medium version. Dictating a recap in your own voice is roughly a thousand times easier than re-outlining from slides.
Reader replies and follow-ups
A thoughtful reply to a commenter is often the seed of the next post. Speaking the reply in full, instead of truncating it to what you have time to type, gives you better drafts and better comments.
The essay you've been threatening to write
Everyone has one — the take you've described at three different dinner parties but never sat down to write. Twenty minutes of dictation is usually enough to finally get it out of your head and onto the page.
Architecture
Drafts belong to you.
So does the audio.
Before a Medium post is published, it's a private document. It may contain company information you haven't cleared with your team, half-formed arguments you're still pressure-testing, and drafts of takes you may or may not actually publish. None of that should pass through a cloud transcription service that keeps logs.
Resonant processes every word on your Mac, using neural speech models tuned for Apple Silicon. Your voice never leaves the device. The only thing that appears in the Medium editor is plain text — the same text you would have typed, just produced at the speed of speech.
This is the architectural line between Resonant and cloud dictation tools like Wispr Flow or Otter. They stream audio to their servers. Resonant doesn't, and can't. Your pre-publication drafts stay between you and your keyboard — now with a voice shortcut.
Free. Local. Works in any text field.
Stop losing the posts
you already wrote in your head.
Voice drafting for Medium writers. No subscription, no cloud audio, no cognitive shift from explaining to performing.
Requires macOS 14+ · Apple Silicon
Related