Your voice is the product. Stop outsourcing it.

By Jack Stovell X · Instagram · LinkedIn · published 2026-08-07 · updated 2026-09-30

Write like you, at the speed everything else moves.

The bind you're in

You are the brand voice. Investors read your updates because they sound like you, not like a comms team. Your LinkedIn works because it's got your fingerprints on it. So you keep writing everything yourself, at 11pm, badly, because delegating means handing your voice to a ghostwriter or an AI tool that flattens it into the same beige executive-speak everyone else sounds like. Fair enough, that's the trade nobody wants to make.

Before you hand anything over, check whether your drafts already sound like AI. We explain the three statistical habits behind that flat, beige feel: sentences of the same length, transitions from one small stock pile, and closing lines that hedge. Once you can see them, you can fix them, and the trade looks less grim.

You don't need a ghostwriter. You need your own voice, measured properly, so it survives the handoff.

What we actually measure

Not vibes. Not "tone: professional, friendly." We run two-pass stylometric analysis across 45 attributes: sentence rhythm, contraction rate, how you open paragraphs, whether you hedge or state things plainly, your actual punctuation habits. Pulled from writing you already have. Old emails. Old posts. That pitch memo you agonised over for three days.

No new writing required. Just the stuff sitting in your drafts folder already.

One profile, every format

Your voice profile doesn't care what you're writing next.

  1. Investor update: same directness, same specificity, none of the "leveraging synergies" nonsense.
  2. LinkedIn post: your rhythm, your fragments, your actual jokes.
  3. Blog piece: longer, still you, not a different person wearing your name.
  4. Launch copy: urgent, on-brand, not generic startup hype.

Build the profile once. Use it everywhere. That's the whole point.

The five-second style check

Every draft comes back with a voice-match score, so you know where to look before you read. A high score means the style has landed, and your read can go on facts and substance. A low score comes with the reason, usually sentence length or a missing contraction pattern, and you fix it fast.

No more re-reading your own AI output six times wondering why it feels off. The score tells you.

Start free, see it work

You can build a full voice profile and run one complete analysis without paying anything. No card, no trial clock ticking down. See your own writing measured before you commit to anything.

If it's useful, upgrade. If it's not, walk away, no harm done.

Plans that scale with how much voice you're managing

Get your voice measured

Free, no card, five minutes.

Still drowning in drafts that don't sound like you? That's exactly the problem we built this for.

The founder writing that matters most

Not everything you write carries the same weight. So stop treating it that way.

Three pieces do the heavy lifting. The investor update, because people read it looking for a reason to trust you or a reason to worry, and your plain sentences decide which. The launch copy, because it's the first thing a stranger hears you say. And the posts under your own name, where readers have learned to recognise your rhythm before they've read a single claim.

Everything else can be drafted by someone else, as long as it's checked. That's the point.

Here's a quick test. Take a line from last quarter's update: "We missed the target. Here's why, and here's what changes." Now picture the delegated version: "We experienced headwinds relative to our objectives." Same facts. Completely different founder. One sounds like a person who'll tell you the truth, the other sounds like a person managing you.

A profile built from your own published writing catches that gap. It knows you state things plainly, that you use fragments, that you hedge rarely. Three or more pieces and about 3,000 words is the recommendation, and you've almost certainly written that already.

A realistic weekly workflow

Forget the grand system. You have one profile and a few hours you can actually spare.

  1. Monday: decide the one piece that matters this week. Just one. Write the rough argument in a few lines, facts only, no polish.
  2. Midweek: draft from that outline. You, a teammate, or an AI tool reading your profile live through the MCP server or the REST API, which covers ChatGPT, Claude, Gemini, Cursor and any MCP client.
  3. Before you read it: check the voice-match score. On-voice drafts typically land 0.85 to 0.95.
  4. Thursday: fix what the score flags, then read for substance.
  5. Friday: publish, and save your edits so the profile learns from them.

The score is the useful bit. A 0.7 doesn't just tell you something's off; the per-feature deltas show the draft's value against yours. Average sentence length too long, say, or contractions missing. That's a two-minute fix, not a rewrite.

Step 3 is where you stop re-reading drafts six times wondering why they feel wrong. You look at the number, look at the reason, and go straight to the paragraph that's drifted.

When the team writes as you

Sooner or later someone else drafts. A marketing hire, a co-founder, a contractor. Fair enough; you can't write every line forever.

The profile is what makes that safe. Your team member drafts, and the draft is scored against your own writing. Habits that slipped get flagged: the hedging that crept in, the sentences that ran long, the opener you'd never use. Partial matches tend to land between 0.5 and 0.8. A clearly different voice comes in under 0.45, which tells you quickly that the draft needs a proper rework, not a light edit.

If you'd rather not send people back and forth, Polish revises a draft toward a target score. The default is 0.9. You can raise it or lower it depending on how much the piece matters.

Two practical notes. First, brief your team on substance, not style. Give them the facts and the point; the profile handles the voice. Second, let them see the deltas. A teammate who learns that they over-qualify every claim will stop doing it by the third draft, and you'll stop being the bottleneck.

To be fair, none of this replaces your judgement on what to say. It just removes the tax of how to say it.

At the end of the day, the goal is simple. Your updates still sound like you, even when you didn't type every word.