Read the whole methodology before you trial.
Four parts, one rubric. What each does, why it’s shaped the way it is, and what we decided not to build. Read it end to end. The trial will make more sense once you have.
Voice goes in. Posts come out.
Most LinkedIn AI tools work in one direction: prompt → draft → ask the user to fix what doesn’t sound like them. The voice approximation is the last step, applied at the surface.
Typelab inverts that order. Voice tuning is the first step. Five samples of your actual writing — an old blog post, a Slack rant, an email you’re proud of — produce a voice profile that captures both what you say and what you’d never say. Every draft, comment, and seed downstream flows through that profile.
The four parts of the product are the workflow that profile enables: drafting in your voice, commenting in your voice on the accounts you want to reach, weekly voice check-ins that seed next week’s posts, and a curated engagement network that activates in the first ninety minutes after you publish.
Voice Tuning
Five samples produce a voice profile.
The voice profile is a structured extraction of how you actually write. Five samples of 200–800 words each are enough — we’ve tested with samples as short as 60 words and as long as 4,000, and the sweet spot is 200–800. You can paste them, upload .docx files, or import from Notion. Slack rants count.
What gets extracted: your archetype (the kind of writer you are — warm operator, dry analyst, restless founder), your signature phrases (the constructions you reach for), the rhetorical moves you make (where you start a paragraph, how you transition, how you close), and the moves you would never make (the stock SaaS language you avoid, the over-claims you don’t make).
The negative-space half is the part most LinkedIn AI tools skip. A voice profile that knows what you would never say is what keeps drafts from sounding like AI — because the model is constrained to stay inside your repertoire, not just to match it.
You can update the voice profile any time. Most users don’t after the first month — the profile’s already there. Some users tune it after a quarter once they’ve seen what consistently feels off.
Post Composer + CGOVE Rubric
Drafts in your voice, scored on a public rubric.
The composer starts from a seed: a thought, a one-liner, a Notion doc, a voice memo, or a transcript from this week’s Voice Check-in. Typelab asks the discovery questions a thoughtful editor would — what do you actually mean by that, what example would land it, what’s the most contrarian version of this take you can defend. Then it drafts.
Every draft is scored against the CGOVE rubric: Clarity, Curiosity gap, Originality, Value, Emotion. Each axis scored 0–20, total out of 100. Below 70: queued back for revision. 70–85: presented to you with a recommendation to ship or rework. 85+: marked ready.
The CGOVE rubric is published in full at /about/editorial. You can grade your own posts against it by hand, with a pen, on paper. We don’t hide the methodology behind a black-box quality score — it’s a learnable, defensible standard you can use without us.
Drafts include hook variants (3 on Solo, 5 on Studio and Executive), marginalia notes (paragraph-level editor's comments on Studio+), and a public interview share link (a permalink to the seed-and-discovery-questions transcript that produced the post, for the readers who want to see the work).
Aspirational Comments
Show up in the right rooms, in your own voice.
Pick five to twenty-five accounts whose audience you’d like to reach. Founders in your category, customers you wish you had, executives whose teams you want to recruit from. Typelab watches their feeds and drafts comments on their fresh posts — in your voice, not theirs.
The drafts go into a manual review queue. You see the post, the proposed comment, and the CGOVE score for the comment. You approve, edit, or skip. The draft is published from your own browser via the Chrome extension — no third-party automation, no anomalous engagement patterns from your account.
Auto-post mode is opt-in and off by default. Even when on, it has a manual approval window before each comment goes live. We are deliberate about this because account safety on LinkedIn is real and we’ve been watching it for eighteen months across 1,200+ professional accounts. The right risk profile is the one you choose, not the one we default you into.
Comments queued: up to twenty per week. The drafts that don’t feel right stay in the queue. Nothing publishes that you haven’t signed off on.
Voice Check-ins
A weekly call that does the hard part of writing.
Once a week or every other week, an ElevenLabs voice agent calls you for ten or fifteen minutes. The cadence is yours: Tuesday morning, Friday afternoon, biweekly, monthly. The agent has your voice profile, your previous transcripts, your queued seeds, and the questions a thoughtful editor would ask given what you’ve been writing about lately.
It asks open questions. You talk. The transcript becomes seeds in your composer queue — three or four to choose from, ready to draft. Most users find the call is the difference between writing five posts a month and twenty.
The agent doesn’t pitch you. It doesn’t try to sell you anything. It doesn’t store your transcripts longer than ninety days unless you opt in to longer retention for cross-week thread continuity. And you can decline a call any week with a single SMS — the call doesn’t fire if you didn’t confirm at 8am that morning.
On Executive, you also get on-demand calls — press a button when you have a take and want to record it before it leaves your head.
The first ninety minutes decide the next nine days.
The Boost Network is a curated set of professional LinkedIn accounts in adjacent spaces. When a member publishes, the network engages in the ninety-minute window where the algorithm is still scoring — the way colleagues in the same building might. Members are anonymous to each other. There’s no reciprocity expected.
We’re rolling Boost Network access out to Studio members through 2026 as the network scales. Trial users join the waitlist and receive an invitation as capacity opens. Read the dedicated page for the full mechanism and operational background.
Decisions we’ve refused to make.
No hook libraries. Hook templates produce posts that sound like every other post that used the template. Voice Tuning replaces the hook library with your own repertoire of openers.
No engagement pods. Pods require reciprocity, draw attention from LinkedIn’s anti-spam systems, and produce the kind of engagement that doesn’t convert. The Boost Network is curated, anonymous, and one-directional.
No third-party post automation. We don’t auto-publish via the LinkedIn API or via account-impersonating browser drivers. You publish from your own browser, on your own cadence, in a way that looks indistinguishable from how you’d post by hand.
No black-box quality score. CGOVE is published in full and learnable on paper. Every draft tells you which axes it’s strong on and which it’s weak on, with the same scoring you can apply yourself.
No content factory. The volume cap on Solo is 30 drafts a month not because we couldn’t serve more — we could — but because the right amount of LinkedIn for most founders is two to four good posts a week, not twenty mediocre ones. Studio is unlimited because some operators do post twenty times a week, and we want them to.
Now you’ve read it. Try it.
Two weeks. No credit card. Full Studio surface, Solo-tier caps. You’ll know within four drafts whether it works for how you write.