Author First: AI Assisted | Steve Castledine | Working v5, 8 October 2026 | Review date: 9 October 2026
Scope: complete supplied practical non-fiction manuscript, including front matter, eleven chapters, six appendices, sources and closing note.
This review offers Casover's editorial assessment of your manuscript. It identifies strengths and explains where, in our judgement, changes could improve the reading experience. Recommendations are supported by examples and reasons, but editorial judgement allows for different interpretations. You decide which suggestions serve the book you want to write. You may adopt, adapt or decline them.
Your book provides a usable route from deciding what help an author wants to checking the work and preparing it for publication. Its strongest quality is the connection between plain advice and decisions readers can actually make. Chapter Five shows how to recognise an edit that damages a claim; Chapter Eight gives a verification method; the appendices supply the working fields rather than merely promising them. The distinction between generated prose, editing assistance and ideas-only work remains consistent across the book.
The evidence does not justify extensive rewriting. The main local repair is to reunite Chapter Seven's voice-profile explanation, which is interrupted by a separate discussion of practising unaided decisions (MC-01). Two references to “supplied” external guides also need to make sense for readers who receive only the book (MC-02).
Publication readiness additionally depends on checks this assessment cannot settle: the truth of dated verification statements and consequential policy, legal, research and real-person claims (MC-03), and usable sources and navigation in the eventual print and ebook files (MC-04). These are specific clearance and production tasks, not evidence that the structure is broken.
Start with the local text repairs, retain the existing chapters and reference material, then close the publication checks. MC identifies findings in this Complete Manuscript Review. The Final Editorial Summary sets the reconciled order across the package.
Title and author: Author First: AI Assisted by Steve Castledine, as stated in the manuscript.
Exact source: Author First revised v5 WORKING - 8 October 2026(1).docx.
Calculated extent: 49,147 whitespace-separated words in the DOCX body paragraphs, including headings, templates and front/back matter. No author-supplied word count. Eleven actual chapters and six actual appendices, plus Sources and Further Reading and Thank You. All supplied body text processed.
Form and subject: Practical non-fiction guide to writing, responsible AI use, self-publishing and book marketing. Its explanatory chapters, worked briefs, adaptable templates and checklists establish the form. Fiction length labels do not apply.
Readership: Manuscript-derived adult fiction and non-fiction authors, especially readers new to AI or cautious about using it. The introduction also supports readers with an existing draft. No separately supplied demographic brief or secondary category; provisional secondary subject is self-publishing practice.
Objective: You supplied publication readiness and minimum justified corrections as this commission's purpose. The manuscript's reader promise is useful, checkable AI assistance with author judgement retained.
Revision history: You state that substantial review and revision have occurred. The manuscript describes an earlier edition and a September rewrite with selected October checks. These are manuscript statements, not independently verified history. No earlier reviews were used.
The provisional editorial position is local repair and controlled completion, as set out in the Quick overview. There is no demonstrated need to rebuild the book.
The book's central question is practical: how can an author use an assistant without letting fluent output make decisions that belong to the author? That question holds together material which could otherwise become a catalogue of tools, prompts and publishing tips. The title is substantiated by the way the chapters repeatedly return from software to the work itself.