legal, and newsletters, and answer a handful of editor-style questions. From there, a team of dedicated agents takes over: a planner commits to a thesis and structure before drafting begins, a writer drafts in the specific voice the job calls for, a fact-checker verifies every claim against real fetched sources rather than model memory, a line editor removes the patterns that make AI writing recognizable, and an independent reviewer reads the finished piece cold. The company's position is that this division of labor, rather than a single model handling everything in one pass, is what separates publishable writing from fluent filler.

Two design choices define the product. First, voice: rather than a company-wide "brand voice," Ghosts trains writers on a specific person's published work, holding that voice for the length of a piece, and each writer keeps learning from every edit and approval within its project. Second, transparency: users see the understanding the system built from their brief before drafting starts, the exact credit cost of a piece before it runs, what the fact-checker flagged, and the draft being written line by line.

Ghosts serves customers whose names go on their work: agencies managing isolated, voice-trained workspaces per client, founders and executives publishing thought leadership, newsletter writers, PR and reputation firms, political consultants, and law firms on a dedicated tier with additional fact-checking and a privacy layer under which client data is never used for AI training.

The platform runs on a credit-based subscription model starting at $29 per month with a 7-day free trial, with studio, agency, and law firm tiers. Ghosts.app was founded in 2026 by Gregory Graf, a consultant who spent 30 years in SEO, content marketing, and online reputation management for multinational brands and high-profile individuals, and built the platform after concluding that no existing AI writing tool produced work he could put his own name on.