A voice guide built from half an hour of you talking.
Record a guided interview in your browser. Get back a ten-section guide, in your own words, as files Claude, ChatGPT, or Gemini can follow.
Question 11 of 18
What is something everyone in your field says that you think is wrong?
“If you are twenty-six and your boss stands up in the all-hands and says, I do not know if we are going to make it, you do not feel trusted. You update your résumé.”
From the answer to question 11
Section 4 of 10
Stances
Your stances are reversals. You held the popular view, something happened, and now you hold the other one, and you tell the reversal as a story with your own mistake in it. You push back gently on your own field and say "maybe I am" old-fashioned rather than defending the position.
“If you are twenty-six and your boss stands up in the all-hands and says, I do not know if we are going to make it, you do not feel trusted. You update your résumé.”
You draw one hard line, about manipulation, and you draw it with a question first and a direct refusal second.
“Some of my fellow coaches think I am old-fashioned. Maybe I am.”
Half an hour of talking, then a guide you can hand to any assistant.
-
Answer eighteen questions
Each one asks for a story, an opinion, or an explanation, never an adjective about your tone. You record one answer at a time, in your browser, in a minute or two each. Re-record any answer, type instead, or skip.
-
Every answer is transcribed
As each answer finishes, it is uploaded and transcribed. A dropped connection loses one answer, never the interview. The recording is deleted once the transcript exists, unless you choose to keep it.
-
The guide is written from your words
Ten sections, in second person, with your own lines quoted in every one. A check requires at least two of your quotes per section. Nothing is invented; a skipped question stays skipped.
-
You get the files
A Claude Skill folder, a CLAUDE.md snippet, a ChatGPT version, a Gemini instruction, a plain VOICE.md, and a designed PDF. Yours to keep, edit, and install.
Ten sections. Every one quotes you.
-
1. Who you are, in your words
A short bio the guide's reader can rely on.
-
2. Vocabulary
Words and phrases you use, and words you refuse.
-
3. Rhythm
Sentence length, paragraph shape, how you open and close.
-
4. Stances
What you believe and what you push back on, with your lines quoted.
-
5. Stories bank
The anecdotes you told, each with the point it makes.
-
6. How you explain
Your analogies, your examples, the order you put ideas in.
-
7. Tone by situation
A client, a stranger, a peer, a critic.
-
8. Formats
How you write an email, a post, a proposal, a reply.
-
9. Banned list
Phrases, structures, and assistant habits that are not you.
-
10. Examples
Three generic drafts rewritten in your voice, with the changes explained.
Files your assistant takes today.
-
For Claude
A Skill folder, zipped. Upload it once under Customize, Skills, and Claude can use it in any conversation.
-
For Claude Code and any agent
A CLAUDE.md snippet that imports the guide. Drop both files in a project.
-
For ChatGPT
Custom instructions under 1,500 characters, plus a project instructions file and the guide as its knowledge file.
-
For Gemini
An instruction block for a Gem. Google has said Gems move to skills; the install notes say what to do when that happens.
-
For anything else
VOICE.md, the ten-section guide as a plain Markdown file.
-
For you
A designed PDF of the guide, for reading and for handing to a person who writes for you.
For people who draft with help and want it to sound like them.
-
Consultants and coaches
Proposals, client notes, posts, and emails, drafted with an assistant and finished by you. The guide holds the instruction you keep having to repeat.
-
Founders
Investor updates, launch posts, hiring notes, replies to customers. One guide across every tool you use.
-
Writers and newsletter authors
Research and outlines with help, the voice held steady. A guide you own as a file and can edit.
One price, paid once.
The snapshot is free with a verified email: two samples in, one page out. The Longhand guide is $39, once. That covers the interview, the ten sections, every file, the PDF, and fifty rewrites a month. A re-interview is $9: five questions on a topic you name, merged into the guide and every file.
What a guide does, and what it does not.
A guide is as good as the interview. Half an hour of specific answers makes a specific guide. Short answers make a short one, and the page tells you when an answer is thin.
An assistant follows a guide the way a new colleague follows a style sheet: well, and not perfectly. You will still edit. You will edit less.
Nothing is trained on your words. The guide is a document a model reads, and you can change any line of it.
The site makes no claim about detectors. A guide makes a draft sound more like you.
Questions people ask.
Can I type instead of talking?
Yes, for any question or all of them. Spoken answers tend to carry more of your voice, which is why the interview is spoken first.
What happens to the recording?
It is deleted as soon as the transcript exists, unless you choose to keep it for re-interviews. The transcript and the guide stay while your account does.
Which assistant does this work with?
Claude, ChatGPT, Gemini, and anything that takes a Markdown file. Each file is described by what its assistant accepts today.
Is this the same as the NexusMKTG service?
It is the self-serve version of the first step. The service is hours of interviews, a dozen guides, and a trained system Brett runs. If you want that, it is one link away.
Built by someone who interviews people for a living.
Longhand comes from a method Brett Wharton uses at NexusMKTG. There, half an hour of his questions and hours of recorded conversation become a dozen brand guides and a trained Claude system. Longhand is the half-hour, self-serve version of the first step.