By the end you will have a reusable Claude Project that loads your brand voice, rules, and audience personas into every new chat, so drafts start on-brand instead of generic.
Every caption you write starts from a blank doc, so the first ten minutes go to reminding the tool who you are and who you are talking to. A voice project moves that briefing out of every chat and into one place, so drafts open on-brand and your edits are about the idea, not the tone.
It is the foundation the rest of the weekly system stacks on.
This is the free first module of urfired.ai's Content & Marketing path. No signup, no card. The steps note what Claude access you need.
In claude.ai, open Projects in the left sidebar and click New Project. Name it for your brand or channel, like "Acme social voice," not "Project 1," and add a one-line description of what it is for.
Free accounts can hold up to five projects, so dedicate one to voice and keep it. Every chat you start inside it inherits its instructions and files automatically.
Before writing anything, pull three of your best posts, the ones that sounded most like you and actually landed. Add one longer piece you are proud of and a short note of words and phrases you never use.
This is the raw material for the whole project. Do not skip ahead to describing your voice in adjectives; show it with real work.
Inside the project, click Set project instructions. Lead with what Claude must not do (no "in today's fast-paced world," no em dashes, no hashtag stacks unless asked), then your role and the platforms you write for, then your tone in three or four concrete words.
Put the hardest rules first, since earlier lines carry more weight. Keep this to behavior and rules, not background.
Open a chat in the project, paste your three posts, and ask Claude to derive a one-page voice brief: sentence length, rhythm, vocabulary, how you open and close, how you sound when you disagree. Edit it until it is actually true to you.
Save it as a text or PDF file and upload it to the knowledge base using the + on the right of the project page. Keep it to one to three pages so the signal stays dense.
Write a short file describing the two people you actually post for: what they want, what they scroll past, the words that land, the objections they carry. Upload it alongside the voice brief.
This is what turns an on-voice draft into an on-target one, so posts are aimed rather than just well-written.
Add a short, specific list of the tells you want gone: cliches, hedges, corporate filler, and any phrasing that reads as machine-written. Put it in the instructions or as its own short file.
This is the highest-leverage piece, because it directly kills the sameness that makes AI drafts sound identical.
Start a fresh chat in the project and give it one real one-line brief you owe this week. Set the draft next to something you would normally write by hand.
Where it drifts off-voice or off-audience, that gap is a missing rule, not a reason to abandon the setup.
Fix the setup, not the single output. Tighten a rule, add a missed phrase to the kill list, or sharpen a persona, then rerun the same brief.
After roughly three passes the project should start most drafts close enough to edit rather than rewrite. Update it again whenever you catch it repeating a mistake.
Take one real one-line brief you owe this week and build a Claude Project around your own voice to draft it. Create the project, paste in your three best-performing posts, and have Claude derive a one-page voice brief you correct and upload.
Add a two-persona file and a do-not list, then run your brief in a fresh project chat and set the output next to what you would have written. Tighten the instructions from what it missed and rerun until the draft needs edits, not a rewrite.
A Claude Project named for your brand, with custom instructions holding your voice rules and a do-not list, plus a one-page voice brief and a two-persona file in its knowledge base, tested on one real brief so new chats start on-brand automatically.
Module 02: Turn positioning into a month of angles and hooks with ChatGPT (ChatGPT, 18 min)
Free gets you this starter lesson today. Pro unlocks the full 14-module path when checkout opens, no card and no charge until then.
New here? The free 2-minute assessment reads your role and gaps, then matches you to the right path. Get your score