One screen, seven phases.
Onboarding walks you through what happens next. Source acquisition, linguistic fingerprint, cognitive style, thematic DNA, personality, psychology, and voice guidelines. No surprises.
We read your public writing across LinkedIn, Twitter, Reddit, personal blogs, podcasts, and press, then build a voice profile that captures how you actually write. Install it in Claude, or paste it into whichever AI tool you use. From then on, your drafts read like you wrote them. The same engine runs in reverse: profile someone you want to reach, so your outreach matches how that person actually thinks. Upload leads one at a time or in bulk, then generate personalized email and LinkedIn outreach for every profile. Free to try with 3 lead-profile credits; My Voice unlocks at $12/mo.
Em-dashes everywhere. Hedged openings. Three-bullet structures. Tidy, weightless conclusions. Every general-purpose model leaves the same fingerprints. People have learned to recognize them. The moment they do, they discount whatever you put in front of them, even if what you put in front of them is good.
A system prompt with a dozen tone instructions does not fix it. The reason AI sounds like AI is not surface tone, it is structure. Vocabulary, sentence rhythm, the way arguments get built, the topics you return to and the ones you refuse. To get drafts that read like yours, your AI needs a structural model of how you actually write.
We build that model from your public writing across every platform we can reach, then hand it to your AI as a single file. Install it once. Your drafts inherit it from then on. The same model also runs in reverse on the people you write to, so your outreach respects how the reader actually thinks instead of guessing.
The pipeline runs the same whether the subject is you or someone you want to reach. It pulls public LinkedIn artifacts, discovers cross-platform writing via search (Twitter, Reddit, personal blogs, podcasts, press), parses every page it finds, then analyzes the whole picture across seven layers. Each layer streams in front of you as it builds. The final output is a single file your AI installs and applies every time it drafts. Only the seventh phase differs by subject: write-as instructions when the subject is you, write-to instructions when the subject is a lead.
LinkedIn profile, posts, articles, comments, and company context are retrieved directly. DuckDuckGo search discovers Twitter, Reddit, personal blogs, podcasts, and press mentions. Each page is parsed for clean text. Sources land in an inventory table as they return, ranked by signal density.
Vocabulary, sentence structure, punctuation habits, opening and closing patterns. The mechanical layer. This is where most tools stop. We start here.
Reasoning patterns, persuasion blend, engagement style. The shape of the arguments and which kinds get taken seriously.
Recurring topics, values, worldview, the personal-professional blend. What the writing returns to, and what it refuses to touch.
Communication temperament, professional disposition. Direct or diplomatic. High-affect or low. Builder or critic.
Motivational hierarchy, emotional expression, information preferences, trust patterns. The deepest layer. What makes a reader open a message versus delete it.
The synthesis. Write-as instructions for your own voice; write-to instructions for lead profiles. Tactical, specific, ready to install.
Two flows, one pipeline. The self-voice flow turns your public writing into a profile your AI installs and drafts from. The lead profile flow runs the same analysis on someone you want to reach, so your outreach matches how they think. Upload leads individually or in batches of up to 50, then generate personalized email and LinkedIn messages for every completed profile.
Onboarding walks you through what happens next. Source acquisition, linguistic fingerprint, cognitive style, thematic DNA, personality, psychology, and voice guidelines. No surprises.
The My Voice page starts empty. One button starts the pipeline. You paste your LinkedIn URL, confirm your name and company, and the system does the rest.
Your voice profile lives here permanently. View the full analysis, download markdown, copy to clipboard, or rebuild from scratch whenever your voice evolves. Rebuilds are unlimited on every paid plan.
Recent leads, credit balance, filter pills. One click to start the next profile. The dashboard hides nothing and asks for nothing.
The LinkedIn prefix stays pinned. You only type the slug, or paste a full URL and watch it strip itself. The payload preview updates as you type.
LinkedIn profile, posts, articles, comments, and company context. DuckDuckGo search discovers Twitter, Reddit, blogs, and press. Each page parsed. Every source lands in the inventory as it returns, ranked by signal density. Nothing gated, nothing inferred.
Linguistic fingerprint, cognitive style, thematic DNA, personality, psychology, voice guidelines. A tactical summary up top for the operator who skims.
What to say, what to avoid, which openings actually land with this person. Distilled from every layer above into the form an agent can act on.
Drop voice-{slug}.skill.md into Claude or Claude Code and it installs as a permanent Skill. Paste the same markdown into ChatGPT, Cursor, or Gemini as context. Either way, every draft you generate for this lead inherits the profile.
Most voice tools live inside a dashboard. You log in, read a personality report, and try to remember the takeaways the next time you write something. That model breaks the moment your AI is doing the drafting.
Every ProvenOutreach profile exports as a SKILL.md file. That is a small markdown document with installation instructions in the header. Drop it into Claude or Claude Code and it installs as a permanent Skill, applied automatically every time the agent drafts for the matching subject. For ChatGPT, Cursor, Gemini, and other tools, paste the markdown as a system prompt or pinned context. Same content, same effect.
Not a template. Not a prompt you re-paste every conversation. A persistent voice profile your stack reuses every time it writes.
--- name: voice-nico-self description: Voice profile for Nico Lafakis's own writing. Apply when drafting LinkedIn posts, tweets, Reddit replies, articles, or any communication being written AS this person. --- # Voice Profile: Nico Lafakis ## Linguistic Fingerprint - Vocabulary: operator lexicon, no buzzwords. - Sentence structure: short, then medium. - Punctuation: no em-dashes. Commas instead. - Openings: declarative claim, not question. ## Application Instructions When drafting content as this person, apply the Voice Guidelines above. Mirror sentence structure and punctuation. Deploy signature phrases when natural, never forced.
Every account starts with 3 free lead-profile credits. Upgrade to Voice to unlock My Voice, or choose a tier with lead credits included.
Try 3 lead profiles on us. My Voice (your own voice profile) requires a paid plan.
Get started free >My Voice profile with unlimited rebuilds. For anyone whose AI drafts in their own name.
Get started >My Voice plus 10 lead profiles per month. For founders writing in their own voice with light outreach.
Get started >My Voice plus 50 lead profiles per month. For operators whose agents draft as them and outbound at scale.
Get started >My Voice plus 250 lead profiles per month. For BD teams and agencies where every agent shares the same voice.
Get started >