Less than a ghostwriter.
Built for the cadence.

Each agent in Pennkin reads your archive — articles, ebooks, transcribed talks, podcast appearances — and learns the patterns of how you think and write. Drafts are grounded in that agent's voice signal, not in generic LLM tone-of-voice prompts. The test is simple: a reader who knows your writing should not be able to tell which posts you wrote and which the agent drafted.

Text-only corpora work too — a handful of substantive past articles is enough to build a meaningful agent. The audio / video ingestion exists because most operators have richer archives there than they realize, not because they're required.

LinkedIn long-form posts and X threads from day one — Pennkin drafts one canonical article and generates a channel-appropriate variant for each, so the LinkedIn version reads like a LinkedIn post and the X version reads like a thread. Substack ships next, followed by Medium and other channels. Every draft goes through your approval before publishing — Pennkin never posts on its own.

It is a ghostwriter alternative for the use case where you want to publish weekly without spending $5k–15k/month or onboarding a human. For longer pieces (essays, op-eds, book chapters), a human ghostwriter is still the right move. For LinkedIn / X cadence, Pennkin replaces the workflow.

Tier prices cover normal usage. Heavy users (very long corpora, frequent re-ingestion, deep research) pass through their incremental LLM cost via our credit metering — transparent, line-itemed, never surprise. Most users never hit the threshold.

First month is full refund, no questions, if voice fidelity doesn't land. We don't ship if the product doesn't pass our own voice test, so refunds are rare — but the offer stands.