A voice lives in two places: how someone sounds, and how they think they sound. The gap is why drafts built from a client's exact words still get “this doesn't sound like me.” Voiceprint measures both in a 15-minute structured discovery, then builds a skill file, custom instructions, and a universal system prompt, so any AI tool you draft in writes on-voice.
You send the draft. It's built from their interviews, their posts, their own phrases. The client reads it, something is off, and they can't tell you what. So you guess. You revise. Round three arrives and the feedback is the same sentence with less patience behind it. Your tone-of-voice doc doesn't settle anything, because a list of adjectives can't be checked against a sentence. The feedback is unfalsifiable. That's the spiral.
The newer version stings more. You draft with AI, the client says it sounds like AI, and now you're defending your process instead of billing for it.
“The copy still ends up sounding like me.”
“This doesn't sound like me.”
“It sounds like AI.”
“Pure AI slop.”
“It uses words I wouldn't use.”
“Client accused me of using AI.”
Voiceprint closes the gap between drafts built from their own words and 'this doesn't sound like me.'
Your client gets a clean, professional discovery. No login, nothing to install. Capturing your own voice: same link, same 15 minutes.
Mostly picks from fixed options, plus short fields for the phrases only they can supply. The picks capture what self-description can't: the client who claims “formal” but chooses the contraction every single time.
A skill file, custom instructions, and a universal system prompt, generated from the answers. Load one into any AI tool and draft. The output holds their voice and cadence, and skips the AI tells.
A structured discovery. It captures cadence, formality, humor tolerance, opinion strength, sentence economy, and the moves a client never wants made in their name.
A skill file, custom instructions, and a universal system prompt, generated from the answers. Works in any AI tool. Plain text, editable, yours.
Every export ships with an AI-tell blocklist baked in: the contrast framings, the tidy triads, the words that trip a reader's alarm. The model gets told what it may never do in this voice.
What the client says they sound like, measured against how they actually choose. Every gap flagged in writing, before you draft a word.
A discovery you'd be comfortable sending at kickoff of a four-figure engagement. Your name on the process, not a chatbot hack.
A ready-to-send kickoff email plus a one-page SOP, so the voiceprint slots into your client process on day one instead of sitting in a folder.
Run one for every client you close. No seats, no subscription, no monthly fee.
Run the discovery on yourself first. Your content, your newsletter, your scripts, your responses, in your own voice.
Done-for-you voice guides start around $4,000. The best-known voice course costs $99 and assigns ten modules of homework before you have anything usable. Voice SaaS runs $59 a month and keeps the voice locked inside its own editor. Voiceprint is $67 once, and the file is yours.
Every client carries an idea of how they sound. Professional, friendly, approachable. Then there's the voice in their existing work: blunt, technical, allergic to small talk. Both are real to them. The distance between the two is the perception gap, and it's where the revision spiral starts. Your draft matched the evidence. They judged it against their self-perception.
Voiceprint measures both sides. The discovery asks what the client thinks they sound like, then runs forced-choice questions that show how they actually choose, and flags every place the two disagree. The gaps are on paper before you draft, so when feedback arrives you have something to point at that isn't your own opinion.
You can, and you'll get an imitation of their past writing. Samples tell a model how someone has written, and nothing about preferences, boundaries, or banned moves. If the back catalog is thin, corporate, or ghostwritten, it's worse: the model imitates writing the client already wants to escape.
And the workflow fails at the exact moment you need it. You cannot onboard a paying client with “paste your posts into the chat.” Voiceprint captures chosen voice through forced choices, works with zero samples, and hands the client a professional discovery instead of a hack.
Most do, and they're fine for your own voice inside that one app. The limits show up in client work: you can't send a built-in style to a client to complete, the voice stays locked inside a single tool, and there's no structured result to read, edit, or put in front of the client for sign-off. Voiceprint's output is a plain, inspectable file you own, and it loads anywhere.
No. $67 once. Every client voiceprint you ever run is included. No monthly fee, and no account required.
All of them. The pack exports a skill file, custom instructions, and a universal system prompt: Claude, ChatGPT, Gemini, Perplexity, anything that takes instructions. Plain text files. Nothing proprietary.
The voiceprint is your paper trail. It's a documented voice spec the client signed off on before the first draft, so feedback gets calibrated against the spec instead of against vibes. “This sounds like AI” turns into a conversation about specific dimensions you can both point at.
Answer seven questions. Voiceprint Mini reads the voice, returns an archetype card, and sends the written breakdown to your inbox.
Run the Voiceprint Mini →