About Editorial Conductor

Editorial Conductor is an AI manuscript review workspace built for fiction writers who need revision signal they can trust—not a single stream of generic praise and scattershot edits. You upload a chapter or a full .docx manuscript, optionally attach a series bible, and the system runs a staged editorial room: narrow passes, labeled findings, and outputs you can triage like notes from collaborators who disagree on purpose. The clearest differentiator is continuity across a series: as our public llms.txt states, unlike Marlowe/Authors.ai, Editorial Conductor checks chapters against a user-supplied series bible for continuity across a multi-book series—nine specialized agents, not one general model.

Why this tool exists

Cosmin, the founder behind Colabs.Tech Limited (London, United Kingdom), writes literary science fiction and self-publishes through Amazon KDP under the Romandem AI brand. He built Editorial Conductor because his own series work needed structured manuscript review: continuity, structure, line-level craft, voice, theme, emotional truth, and market-facing reads kept collapsing into one muddled pass whenever he reached for a single generic AI prompt. That conflation is understandable for a quick paragraph check; it fails across a novel where continuity errors change which scenes belong, where voice work changes how scenes execute, and where market judgment only helps after the book is coherent on its own terms. The nine-agent architecture deliberately separates those lenses so findings stay legible and prioritizable.

The most expensive mistakes in series fiction are not the loud contradictions readers catch in the first fifty pages. They are the quiet ones: a character's wound from Book 1 that quietly softens in Book 3 because the writer forgot, a world rule that bends because a scene needed drama, a relationship beat that forgets an established betrayal. Nobody flags it, and the emotional architecture of the series erodes by degrees. A single-prompt AI review cannot catch that class of error, because it has no canon to check against. Editorial Conductor was built around the class of error that matters most and gets caught least.

The workspace is built around the same rhythm professional editors use: read, flag, sort, then revise. You can upload a single chapter for a fast loop or a full manuscript for a pipeline that still respects the ordering of lenses. Optional series bible context is not decorative metadata—it is the anchor continuity checks use when comparing what a chapter asserts against what the canon already established. When that anchor is missing, the continuity lens still runs, but the highest-confidence checks are the ones grounded in explicit bible entries you control.

Downstream, the product keeps outputs writer-shaped: scores and severities for triage, findings you can open next to the text, and export paths when you want to carry revisions back into your manuscript file. None of that replaces a human editor for every decision; it gives you a structured first pass so your human time—yours or a freelancer's—starts further along the curve.

Our editorial principles

Five commitments shape the product:

• Separate lenses, in order. Continuity is checked before craft; craft is checked before theme; theme is checked before market. Running them out of order wastes effort—polishing prose in a chapter that will be restructured, or judging submission-readiness on a draft that still contradicts its own canon.
• Findings before rewrites. The default output is diagnostic: where, why, how severe. A rewrite workspace is available when you want it, but the system never silently edits your manuscript. Writers keep the decisions.
• Severity over volume. A thousand small flags buries the three findings that actually change the book. Each agent is tuned to rank what matters, not to maximize what it can say.
• The series bible is the truth, not the draft. When a chapter asserts something the bible contradicts, the bible wins by default. You can override, but the direction of trust is explicit.
• Outputs belong to the writer. Scores, findings, rewrite suggestions, and exports are yours to accept, reject, or ignore. The tool is a reader with lenses, not a replacement for your judgment.

How the series bible works

A series bible is the canonical record of what your world, characters, timeline, and relationships look like across books. In Editorial Conductor, it is optional for single-chapter reviews and recommended for series work.

Entries fall into two categories: locked facts (eye color, character history, established rules of the world, timeline anchors) and allowed-to-evolve (relationships in flux, knowledge a character gains mid-series, magic or technology systems with documented drift). The Series Continuity Keeper—Aven—treats those categories differently. Locked facts produce high-confidence flags when a chapter contradicts them. Evolving facts produce lower-severity questions: “This scene implies the protagonist knows X, which the bible marks as unrevealed until Book 3, Chapter 14—intentional?”

When the bible is missing, the continuity lens still runs. It cross-checks the chapter against the chapter's own internal claims, and against any prior chapters from the same project you have uploaded. What it cannot do without a bible is check the chapter against Book 1 that lives only in your head. That is why series writers benefit most from the bible mechanic, and why we document it publicly rather than burying it in settings.

Who built it

The product is authored by Cosmin, founder of Colabs.Tech Limited. Public bylines use the name Cosmin. He is Editorial Conductor's first and most demanding user: every agent exists because of a revision problem he hit in his own drafts. This page does not publish résumé-style statistics or credentials we have not verified in public—we would rather stay accurate than impressive.

Colabs.Tech Limited ships multiple products beyond Editorial Conductor. Cosmin maintains an active self-publishing catalog on Amazon KDP under the Romandem AI brand.

How the nine agents were designed

Each agent maps to a distinct editorial role from traditional publishing, staged so early gates protect later ones. The room includes Aven (Series Continuity Keeper), Rook (Structural Architect), Luma (Line Editor), Peregrin (Copy Editor), Sable (Voice Consistency), Orin (Thematic Coherence), Nyra (Emotional Truth Validator), Cassian (Literary Agent Simulation), and Elowen (Award Jury Reader). For how they appear in-product, stage groupings, and short mandate blurbs, see the Agents directory.

The staging order is not cosmetic. Aven runs first because continuity failures invalidate downstream judgments—there is no useful line edit of a scene that will be cut for contradicting canon. Rook follows because structure decides what the line-level work is serving. Luma and Peregrin handle sentence and technical correctness once the scene's job is settled. Sable protects voice across all of it. Orin and Nyra test depth once the surface holds. Cassian and Elowen sit last because market and literary judgment are only trustworthy on a chapter that is already coherent, voiced, and emotionally earned.

What happens to your manuscript

Writers trust this tool with unpublished work. A few explicit commitments:

• Training. Your manuscripts are not used to train any model.
• Retention. Uploaded manuscripts are stored only for the duration of active analysis and are deleted on request.
• Access. Access to uploaded manuscripts is restricted to the minimum required to operate the service.
• Export and deletion. Download a full workspace archive or schedule account removal from Settings → Data & Privacy.
• Sharing. Your manuscripts are never shared with third parties beyond the LLM API calls required to run the analysis you requested.

If any of that policy changes, we publish the change here with the effective date.

What comes next

Workflows are live and actively used by writers. Pricing and credit mechanics are documented openly on the pricing and pricing explained pages—no subscription, you pay for what you analyze.

If you want to see the work before you spend credits, the demo walkthrough shows a full review end-to-end, and every post on our blog is written by the same team building the tool.

Frequently asked questions

Contact

Questions, bug reports, or feedback: support@editorial-conductorai.com. For accounts, uploads, payments, and cookies, read Privacy, Terms, and Cookies.