Author First: AI Assisted | Steve Castledine | Working v5, 8 October 2026 | Review date: 9 October 2026
Scope: ideas, argument, instructional emphasis and internal support in practical non-fiction. Complete supplied text processed.
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.
The book's main idea is coherent: useful AI assistance depends on the author defining the work and judging the result. This is developed through decisions rather than left as a slogan. Briefs, working modes, source checks, templates, publishing records and marketing limits each give that responsibility a practical form.
Three productive tensions remain visible. AI can save effort while generating repair work; an author can direct a book without automatically owning copyright in every generated element; and a tool can improve a passage without demonstrating lasting improvement in the writer's skill. The book generally explains these distinctions rather than pretending they disappear.
There is no supported thematic contradiction requiring a new argument or a replacement conclusion. Repeated safeguards are usually attached to a fresh task or reference route. Their unequal emphasis is appropriate: checking and judgement are central to the promise.
TB-01 concerns the evidence boundary. Dated research and policy statements, and statements that sources were checked, must match the actual evidence record before release. This is a verification need, not an established error. Preserve the book's qualifications and the distinction between constructed illustrations and real evidence. TB identifies Thematic Balance findings; use the Final Editorial Summary for the final order of work.
This is an expository practical guide for adult authors, not a symbolic narrative. Its subject is AI-assisted writing and publishing. Its main argumentative principle is author responsibility exercised through clear tasks, scrutiny and decisions.
Steve's supplied objective for this commission is publication readiness with minimum justified corrections. The ideas below are manuscript-derived, not separately confirmed author intentions. The book does not need a fictional theme arc.
| Idea | Type | Role | Evidence locations |
|---|---|---|---|
| Author leads; assistance is chosen | Central argument | Organises the method | Chapters One, Five, Seven and Eleven |
| Evidence before confidence | Supporting argument | Makes answers checkable | Chapters Two and Eight; Appendix F |
| Responsibility, rights and disclosure differ | Conceptual distinction | Prevents false equivalence | Chapter Three; Appendix D |
| Preserve voice and useful writing | Editorial principle | Controls revision | Chapters Five and Eight; Appendix B |
| Manageable work and proportionate cost | Practical principle | Controls workload and tools | Chapters One, Four, Seven and Ten |
| Truthful reader-facing presentation | Practical/ethical principle | Connects manuscript to publication | Chapters Nine and Ten; Appendices C and D |
| Idea | Relative emphasis | Development | Status |
|---|---|---|---|
| Author leads | Dominant | Applied throughout | Manuscript-derived argument |
| Evidence | Strong | Citations, claims, research and changing rules | Manuscript-derived argument |
| Rights/disclosure distinction | Concentrated then reinforced | Before drafting and at publication | Explicit explanation |
| Voice | Strong in writing/editing | Samples, choices and profiles | Manuscript-derived principle |
| Manageable work | Recurring | Session to tool and marketing decisions | Explicit practical advice |
| Truthful presentation | Increasing towards publication | Copy, reviews, declarations and updates | Explicit practical advice |
These are ideas and teaching principles, not invented motifs. No requirement exists for equal word allocation.
Chapter One starts with purpose and the next task. Chapter Two establishes the fallibility of fluent answers. Chapter Three complicates responsibility by separating it from rights and platform acceptance. Chapter Four translates judgement into task-based tool tests and privacy questions.