★ Flagship Guide · Creative Writing Multi-AI
June 5, 2026 · 20-min read
I was 80K words in when my protagonist’s eye color changed three times. So I built 10 AI workflows that never let it happen again.
Scattered notes, contradictory timelines, forgotten subplots, maps that don’t match the text — every novelist hits the same wall. These 10 multi-agent workflows turn NotebookLM + Claude + ChatGPT + Gemini into your personal continuity department.
One Master Prompt runs the system. Ten specialized workflows handle every creative writing problem. Here’s the free one that starts it all.
⚡ Featured Prompt of the Day — Your First Win in 30 Seconds
You are now the Creative Writing AI Command Center. We have 4 specialized agents working in strict sequence. Roles and rules:
(1) ARCHIVIST [NotebookLM] — Only extracts raw data from uploaded sources. Never interprets, summarizes, or invents. Every fact must have a source citation. Output format: bullet list with [SOURCE: document name, page/chapter].
(2) ARCHITECT [Claude/ChatGPT] — Only structures the Archivist's extracted data into organized systems (database entries, tree diagrams, timelines, maps). Never adds new information. Never resolves conflicts — flags them with [CONFLICT].
(3) AUDITOR [ChatGPT/Claude] — Only checks the Architect's structured output against the original sources for: contradictions, missing data, logical impossibilities, continuity errors. Never fixes — only flags with [SEVERITY: Critical/Major/Minor] and exact source references.
(4) WEAVER [Gemini/Claude] — Only produces the final clean deliverable by resolving all [CONFLICT] and [SEVERITY] flags from the Auditor. Every resolution must cite which source version was chosen and why.
Process (strict order):
Phase 1: ARCHIVIST extracts from [UPLOAD YOUR MANUSCRIPT/NOTES]
Phase 2: ARCHITECT structures into [SPECIFY: character database / timeline / family tree / relationship map / worldbuilding index / story bible / continuity report / plot hole audit / entity tracker / location map]
Phase 3: AUDITOR flags all issues
Phase 4: WEAVER produces final clean output
Task: [DESCRIBE YOUR SPECIFIC NEED]
Begin Phase 1 now.
Updated June 5, 2026 · Battle-tested on 120K-word manuscripts with 80+ characters · 100+ iterations · 1 Master Prompt free — full system in the Creative Writing Collection
TL;DR — 10 multi-agent workflows for novelists and screenwriters: character databases, family trees, timelines, continuity checks, relationship maps, worldbuilding indexes, plot hole detectors, interactive maps, story bibles, and entity trackers. Each workflow uses NotebookLM + Claude + ChatGPT + Gemini in a Perceive → Plan → Act → Evaluate loop. Free Master Prompt included.
Updated June 2026. Maintained by a small team of AI super-users who teach multi-AI research and creative writing workflows — no affiliate relationships. About this guide →
Why a single AI was silently destroying my manuscript
For two years I did what most novelists do — asked one AI to “check my manuscript for errors.” It would find the obvious stuff: typos, repeated words, basic grammar. But it never caught the protagonist’s eye color shifting from blue to green to hazel across 47 chapters. It never noticed that a character who died in chapter 12 appeared alive in chapter 31. It never flagged that the travel time between two cities was three days in one chapter and three hours in another.
The problem wasn’t the AI’s capability. It was the architecture. One prompt doing everything means nothing gets done sharply. So I built what I call the Command Center: four specialized agents, each locked into one job, working in strict sequence.
The Archivist only extracts. The Architect only structures. The Auditor only flags. The Weaver only resolves. No agent is allowed to do another agent’s job. That constraint is what makes the system work — the same way a real editorial department works, where the copyeditor doesn’t rewrite your prose and the fact-checker doesn’t make structural suggestions.
★ Why it works
Specialized agents in strict sequence catch what one AI smooths over.
10Workflows
4AI tools
12minAvg per workflow
- Locked roles prevent mush. The Archivist can’t interpret. The Auditor can’t fix. Each output is pure — and that purity is what makes contradictions visible.
- NotebookLM grounds everything in your actual manuscript. No hallucinated characters, no invented plot points. Every fact is cited to a source document.
- The Auditor phase is where the magic happens. That’s where the eye-color changes surface, where the dead character reappears, where the impossible travel times get flagged. Without a dedicated auditor, these hide in plain sight.
The 10 workflows that cover every creative writing problem
WF 1 — Character Database
Structured profiles for every character in your manuscript
Extracts names, ages, appearances, motivations, speech patterns, and relationships from scattered notes. One source of truth for your entire cast.
WF 2 — Family Trees
Auto-generated genealogy with generation tracking
Maps blood lines, marriages, adoptions, and found families across generations. Flags impossible age gaps and contradictory relationships.
WF 3 — Timeline Tracking
Master timeline with age calculations at every event
Orders events chronologically, calculates character ages at each point, flags temporal paradoxes (two places at once, impossible travel).
WF 4 — Continuity Checking
Every inconsistency caught before your readers find it
Scans for eye-color changes, name spelling shifts, timeline contradictions, object ownership errors, and established-rule violations.
WF 5 — Relationship Maps
Who knows who, who hates who, who loves who
Maps the emotional web between all characters: trust, resentment, power dynamics, key turning points, and current status.
WF 6 — Worldbuilding/Lore
One index for your entire fictional world
Systematizes magic systems, political structures, religions, economies, languages, and world history. Flags rule violations.
WF 7 — Plot Hole Detection
Every logical gap found before your editor does
Checks cause-effect chains, character motivation consistency, unresolved setups (Chekhov’s guns), and convenience problems.
WF 8 — Interactive Maps
Locations, distances, travel routes, settlements
Extracts geography from text, builds spatial relationships, calculates travel times, flags geography contradictions.
WF 9 — Story Bible
One reference document from 200 pages of scattered notes
Compiles premise, themes, character briefs, world overview, chapter summaries, tone guide, and unresolved questions.
WF 10 — Entity Tracker
Every creature, mount, weapon, artifact, and important object
Catalogs non-human entities with ownership history, power changes, appearances, and status tracking. See details ↓
The Perceive → Plan → Act → Evaluate loop
P
Perceive
Archivist mines your sources. Every fact cited. No interpretation.
L
Plan
Architect designs the output structure. Database schema, tree format, timeline layout.
A
Act
Architect populates the structure. Auditor flags every issue.
E
Evaluate
Weaver resolves all flags. Final clean output with cited resolutions.
Workflow 1 — Character Database Generation & Maintenance
Perceive
Archivist mines all sources for character mentions
Plan
Architect designs 10-field profile schema
Act
Architect populates profiles. Auditor flags conflicts.
Evaluate
Weaver resolves conflicts, outputs clean database
Agent Roles
Archivist
NotebookLM
Mines uploaded manuscripts and notes. Extracts every character mention with chapter/page citation. Never interprets personality — only quotes direct descriptions.
Architect
Claude
Designs and populates the 10-field character profile template. Structures raw extractions into consistent database entries. Flags conflicting data with [CONFLICT] tags.
Auditor
ChatGPT
Cross-references every profile field against all source documents. Checks for: missing data, contradictions, inconsistent spellings, impossible ages, and duplicate entries.
Weaver
Gemini / Claude
Resolves all flagged conflicts by choosing the most recent or most detailed source version. Produces the final clean character database with resolution notes.
Copy-Paste Prompts
You are the ARCHIVIST. I am uploading my manuscript and character notes as source documents.
Your job: Extract every character mention from these sources. For each character, pull:
- Every name variant (full name, nickname, title, misspelling)
- Every physical description (exact quote + source)
- Every stated or implied age/birth year
- Every stated personality trait (exact quote + source)
- Every relationship mentioned (who, what type, source)
- Every location associated with them
- First appearance (chapter/page)
Rules:
- Never interpret. Only extract verbatim.
- Never merge characters — if two mentions might be the same person, flag as [POSSIBLE DUPLICATE: Name A vs Name B]
- If a description contradicts another, list BOTH with full citations: [CONFLICT: Source A says X, Source B says Y]
- Output as a structured bullet list grouped by character name
Begin extraction.
You are the ARCHITECT. Input: the Archivist's extracted character data.
Your job: Structure this data into a Character Database with these 10 fields per character:
1. NAME — Canonical name + all aliases
2. AGE/BIRTH YEAR — Stated age with source; calculated age if derivable
3. PHYSICAL DESCRIPTION — Consolidated from all mentions (flag contradictions)
4. PERSONALITY — Top 5 traits with source citations
5. CORE MOTIVATION — What drives this character (1 sentence)
6. KEY RELATIONSHIPS — List of [Character Name: Relationship Type]
7. SPEECH PATTERNS — Verbal tics, accent markers, vocabulary level
8. FIRST APPEARANCE — Chapter/page where character first appears
9. CHARACTER ARC — 1-sentence summary of their journey
10. STATUS — Alive / Dead / Unknown (with chapter where status changes)
Rules:
- Never add information not in the Archivist's data
- Never resolve contradictions — mark them [CONFLICT]
- Never merge characters — mark [POSSIBLE DUPLICATE] for human review
- Output as a structured database entry for each character
Populate the database now.
Steps
01
Upload sources to NotebookLM
Put your full manuscript, character sheets, outlines, and any supplementary notes into one NotebookLM notebook. More sources = better extraction. NotebookLM can handle up to 50 sources.
02
Run the Archivist prompt in NotebookLM
Paste the Archivist prompt into NotebookLM’s chat. It will extract character data grounded in your actual sources with citations. Save the output.
03
Run the Architect prompt in Claude
Copy the Archivist’s output into a new Claude conversation. Paste the Architect prompt. Claude will structure everything into the 10-field database format.
04
Run the Auditor prompt in ChatGPT
Take the Architect’s output and paste it into ChatGPT with the Auditor prompt. ChatGPT will cross-reference against the original sources and flag every issue.
05
Run the Weaver prompt for final output
Take the Auditor’s flagged output and run it through the Weaver in Gemini or Claude. The Weaver resolves every conflict and produces the clean, final database.
💡 Token efficiency: Run the Archivist on your manuscript in batches (Part 1, Part 2, etc.) if it exceeds 100K characters. Merge extractions before running the Architect. This prevents context window overflow and keeps extraction quality high.
💡 Robustness: Always keep the raw Archivist output as a permanent record. When you add new chapters, re-run only on the new material and merge with the existing database. Don’t re-extract the entire manuscript each time.
Workflow 2 — Family Tree Creation & Automatic Updates
Perceive
Archivist extracts all familial relationships from text
Plan
Architect designs generation-based tree structure
Act
Populate tree. Auditor checks age gaps and relationships.
Evaluate
Weaver resolves conflicts, outputs clean ASCII tree
Agent Roles
Archivist
NotebookLM
Extracts every family relationship: parent-child, siblings, marriages, adoptions, in-laws. Cites exact source passages.
Architect
Claude
Builds hierarchical tree with generation numbering. Uses ASCII or text-based diagram format. Separates blood lines from marriages.
Auditor
ChatGPT
Checks: impossible age gaps (parent younger than child), contradictory birth orders, characters in two family branches simultaneously.
Weaver
Gemini
Resolves all genealogical conflicts. Produces final tree with generation labels, relationship key, and update instructions.
Copy-Paste Prompt
You are the Creative Writing Family Tree System. 4 agents, strict sequence.
ARCHIVIST: From the uploaded sources, extract ALL family relationships. For each: Person A → Relationship Type → Person B, with exact source citation. Include: parent-child, sibling, marriage, adoption, in-law, foster/step-family, found family. Flag [IMPLIED] for relationships not explicitly stated but strongly suggested.
ARCHITECT: Build a hierarchical family tree from the extracted data. Format as ASCII text tree with:
- Generation 0 (eldest known ancestors) at top
- Blood lines marked with │ and ├──
- Marriages marked with = or + between names
- Adopted/found family in [brackets]
- Question marks for unknown parents
- Each person labeled: Name (age/birth year if known)
AUDITOR: Check the tree for:
- [AGE GAP] Parent appears younger than child
- [BIRTH ORDER] Contradictory statements about eldest/youngest
- [IMPOSSIBLE] Character appears in two unrelated family branches
- [DEAD ALIVE] Character dies but appears in later generation
- [TIMELINE] Marriage or birth dates don't match story timeline
WEAVER: Resolve all flags. Output final clean family tree with resolution notes.
Sources: [PASTE YOUR MANUSCRIPT EXCERPTS OR DESCRIBE YOUR NOTEBOOKLM NOTEBOOK]
Begin.
💡 Token efficiency: Family trees work best when you pre-filter to only family-relevant chapters. Don’t feed action sequences — feed dialogue scenes, backstory chapters, and character introduction scenes.
💡 Maintenance: After each new chapter with family revelations, run the system on just that chapter and ask: “Update the existing tree with these new relationships. Flag any conflicts with the current tree.”
Workflow 3 — Timeline Tracking & Age Calculations
Perceive
Archivist extracts every temporal marker from text
Plan
Architect builds chronological event framework
Act
Calculate ages. Flag paradoxes and impossible sequences.
Evaluate
Weaver resolves, outputs clean timeline with age column
Agent Roles
Archivist
NotebookLM
Extracts every temporal reference: dates, “three months later,” seasons, character age mentions, “the year the war began,” astronomical events, holidays.
Architect
Claude
Converts all temporal references to a unified timeline. Calculates character ages at each event based on stated birth years/ages. Uses relative dating when absolute dates aren’t available.
Auditor
ChatGPT
Flags: temporal paradoxes, characters in two places at once, impossible travel times, contradictory duration statements, age inconsistencies at event points.
Weaver
Gemini
Resolves all temporal conflicts. Produces master timeline table with columns: Event ID, Date/Time, Location, Characters, Ages, Chapter, Causal Links.
Copy-Paste Prompt
You are the Creative Writing Timeline System. 4 agents, strict sequence.
ARCHIVIST: From the uploaded sources, extract EVERY temporal reference. Categories:
- Explicit dates (years, months, days)
- Relative durations (“three weeks later,” “a year had passed”)
- Seasonal markers (“the first snow,” “harvest time”)
- Character age references (“she was twenty-five when...”)
- Historical markers (“ten years after the war”)
- Astronomical/calendar events (solstices, eclipses, festivals)
For each, cite: exact quote, chapter/page, which characters are present.
ARCHITECT: Build a master timeline. For each event:
- Event ID (sequential)
- Estimated date (absolute or relative to anchor event)
- Location
- Characters present (with calculated ages)
- Chapter/page reference
- Causal links to other events (caused by / leads to)
Use the earliest explicit date as your anchor point. Calculate all others relative to it.
AUDITOR: Check for:
- [PARADOX] Character appears in two events at the same time in different locations
- [IMPOSSIBLE TRAVEL] Events imply travel faster than physically possible
- [DURATION CONFLICT] Two passages give different durations for the same interval
- [AGE ERROR] Character’s calculated age doesn’t match a later stated age
- [DEAD APPEARANCE] Character appears at events after their death
WEAVER: Resolve all flags. Output clean timeline table. Include a separate “Age at Event” column for each major character.
Sources: [PASTE EXCERPTS OR DESCRIBE NOTEBOOKLM NOTEBOOK]
Begin.
💡 Token efficiency: For non-linear narratives, process each timeline strand separately first (present-day chapters, flashback chapters, etc.), then run the Architect on the merged extractions to build the unified timeline. This prevents context confusion.
💡 Robustness: Create an “anchor events” sheet: list 5-10 events with confirmed absolute dates. The system calculates everything else relative to these anchors. When you revise, only update the anchors and re-run.
Workflow 4 — Continuity Checking Across Chapters
Perceive
Archivist extracts all physical/object/setting details
Plan
Architect organizes details into cross-referenceable index
Act
Auditor compares every detail across all chapters
Evaluate
Weaver produces fix suggestions for each error
Agent Roles
Archivist
NotebookLM
Extracts every concrete, verifiable detail: physical descriptions, object locations, room layouts, weather, time of day, clothing, injuries, spellings of names/places.
Architect
Claude
Organizes extractions into a master continuity index grouped by category (Characters > Physical, Characters > Objects, Settings > Layouts, etc.) with chapter references.
Auditor
ChatGPT
Systematically compares every detail across chapters. Flags: [PHYSICAL], [OBJECT], [SETTING], [NAME], [TIMELINE], [RULE] errors with exact chapter/line references.
Weaver
Claude
For each flagged error, proposes 2-3 fix options ranked by minimal disruption to the narrative. Suggests which version to keep and what to change.
Copy-Paste Prompt
You are the Creative Writing Continuity System. 4 agents, strict sequence.
ARCHIVIST: From the uploaded manuscript, extract every concrete, verifiable detail. Categories:
- PHYSICAL: eye color, hair color, height, build, scars, tattoos, birthmarks, injuries, clothing
- OBJECTS: who carries what, where objects are, ownership transfers
- SETTINGS: room layouts, building descriptions, landscape features, distances
- NAMES: character names, place names, titles (check all spellings)
- RULES: magic/technology rules, social rules, established facts about the world
For each detail: exact quote, chapter, page/paragraph reference.
ARCHITECT: Build a Continuity Master Index. Structure:
Category → Subcategory → Detail → All mentions (Chapter: exact quote)
Example: Characters > Elena > Eye Color > Ch3: "her deep blue eyes" / Ch21: "those green eyes flashed" / Ch41: "she brushed a strand from her hazel eyes"
AUDITOR: Compare every detail across all its mentions. Flag:
- [PHYSICAL] Contradictory physical descriptions
- [OBJECT] Object in wrong hands or location
- [SETTING] Layout doesn't match previous description
- [NAME] Spelling/naming inconsistency
- [RULE] Established rule violated later
- [DEAD/ALIVE] Status contradiction
For each flag: SEVERITY [Critical/Major/Minor], both conflicting passages with full citations, and chapter order.
WEAVER: For each flagged error, provide:
1. The error summary
2. Option A: Keep the EARLIER version, change the LATER one
3. Option B: Keep the LATER version, change the EARLIER one
4. Recommendation: which option requires fewer downstream changes
Sources: [PASTE MANUSCRIPT OR DESCRIBE NOTEBOOKLM NOTEBOOK]
Begin.
💡 Token efficiency: For manuscripts over 100K words, run continuity checks by Part/Act rather than the full manuscript. First run: Chapters 1-15. Second run: Chapters 16-30. Third run: cross-check the two outputs against each other.
💡 Maintenance: After every revision session, re-run the Archivist on only the changed chapters. Compare the new extractions against your existing Continuity Master Index. This incremental approach saves tokens and catches regression errors.
🔒 Premium · Workflows 5–10
Unlock the remaining 6 workflows with full copy-paste prompts, agent configs, and step-by-step guides.
Relationship Maps · Worldbuilding/Lore Management · Plot Hole Detection · Interactive Maps · Story Bible Generation · Entity Tracker
Each workflow includes 4 specialized agent prompts, the Perceive-Plan-Act-Evaluate pipeline, token optimization tips, and maintenance protocols.
$19.99
Unlock Collection →
One-time payment · Permanent access · Instant download
Workflow 5 — Relationship Maps Between Characters
Perceive
Archivist mines all character interactions
Plan
Architect builds adjacency list with emotional metadata
Act
Map tension arcs. Flag unexplained relationship shifts.
Evaluate
Output relationship web with turning-point index
[LOCKED] Full 4-agent Relationship Map prompt with Archivist, Architect, Auditor, and Weaver roles. Includes: adjacency list format, emotional metadata fields, power-dynamic tracking, turning-point identification, and relationship-arc visualization. Unlock to copy.
Workflow 6 — Worldbuilding / Lore Management
Perceive
Archivist extracts all worldbuilding elements
Plan
Architect systematizes into 8 lore categories
Act
Auditor checks every rule against every usage
Evaluate
Output lore index with rule-violation report
[LOCKED] Full 4-agent Worldbuilding/Lore prompt. 8 extraction categories: Geography, Political Systems, Magic/Technology (with rules and limits), Religion, Culture, History, Economics, Languages. Includes rule-violation detection and cross-reference indexing. Unlock to copy.
Workflow 7 — Plot Hole & Contradiction Detection
Perceive
Archivist maps all plot events and causal links
Plan
Architect builds cause-effect chain diagram
Act
Auditor traces every effect to its cause, flags gaps
Evaluate
Output: plot hole report ranked by severity
[LOCKED] Full 4-agent Plot Hole Detection prompt. Checks: cause-effect chains, character motivation consistency, world-rules consistency, unresolved setups (Chekhov’s guns), convenience problems. Severity-ranked output with fix suggestions. Unlock to copy.
Workflow 8 — Interactive Maps: Locations, Routes & Settlements
Perceive
Archivist extracts all location and travel references
Plan
Architect builds spatial relationship model
Act
Calculate distances. Flag impossible travel times.
Evaluate
Output location index + route table + settlement profiles
[LOCKED] Full 4-agent Interactive Maps prompt. Extracts locations with climate, terrain, features, population, inhabitants. Builds route table with origin, destination, distance, travel time, terrain difficulty, dangers. Flags geography contradictions. Unlock to copy.
Workflow 9 — Story Bible Generation from Notes & Drafts
Perceive
Archivist mines all notes, outlines, and drafts
Plan
Architect structures the story bible template
Act
Fill template. Auditor identifies gaps and contradictions.
Evaluate
Output: complete story bible with gap report
[LOCKED] Full 4-agent Story Bible prompt. Compiles: Premise & Logline, Theme Statement, Character Profiles, World Overview, Chapter-by-Chapter Plot Summary, Tone & Style Guide, Rules of the World, Unresolved Questions. Includes gap detection and cross-reference index. Unlock to copy.
Workflow 10 — Tracking Creatures, Mounts, Objects & Important Items
Perceive
Archivist catalogs all creatures, objects, mounts
Plan
Architect builds entity taxonomy with lifecycle fields
Act
Track status changes. Flag impossible appearances.
Evaluate
Output: entity catalog with lifecycle timeline per item
[LOCKED] Full 4-agent Entity Tracker prompt. Catalogs: CREATURES (name, type, abilities, habitat, behavior, first appearance, status changes), OBJECTS (name, type, powers, owner history, condition changes), MOUNTS (name, species, rider, appearances). Lifecycle timeline per entity with contradiction flags. Unlock to copy.
🔒 Unlock the Full Creative Writing Collection
6 more workflows. 24+ copy-paste agent prompts. Full step-by-step guides.
Relationship Maps · Worldbuilding · Plot Hole Detection · Interactive Maps · Story Bible · Entity Tracker
$19.99
Unlock Collection →
One-time payment · Permanent access · Instant download
Quick reference: which workflow for which problem
| Problem You’re Facing | Run This Workflow | Frequency | Time |
|---|---|---|---|
| “What color are her eyes again?” | WF 4 Continuity Check | After every major revision | ~15 min |
| “How are these two characters related?” | WF 2 Family Tree | Once + updates per new chapter | ~10 min |
| “When did this happen relative to that?” | WF 3 Timeline | Once at midpoint + once at end | ~15 min |
| “I need a quick refresher on Character X” | WF 1 Character Database | Every 5 chapters | ~12 min |
| “Would Character A actually trust Character B?” | WF 5 Relationship Map | Every 10 chapters | ~12 min |
| “Does my magic system still make sense?” | WF 6 Worldbuilding Index | Once at midpoint + once at end | ~20 min |
| “Is there a logical gap in the plot?” | WF 7 Plot Hole Detection | During revision phase | ~15 min |
| “How far is City A from City B?” | WF 8 Interactive Maps | Once + updates for new locations | ~15 min |
| “I need everything in one document” | WF 9 Story Bible | Once + quarterly refresh | ~25 min |
| “Where is the enchanted sword right now?” | WF 10 Entity Tracker | Every 10 chapters | ~12 min |
Frequently asked questions
Which AI tool should I use for creative writing workflows?
NotebookLM for grounding in your actual manuscript. Claude for nuanced character analysis and structuring. ChatGPT for systematic auditing and contradiction detection. Gemini for synthesis and final narrative output. Use them together for best results, but Claude alone can do everything in a pinch.
Can I use just one AI tool instead of four?
Yes. The Master Prompt works in any single tool. You lose the specialization benefit (each tool has different strengths), but the multi-agent architecture still forces sharper output than a single generic prompt. Start with Claude, add others as you see the difference.
How often should I run these workflows?
Character database: after every 5 chapters. Continuity check: after every major revision. Timeline: once at midpoint and once at completion. Plot hole detection: during the revision phase. Worldbuilding: once when your world is built, then after major additions. The Quick Reference table above has specific recommendations.
Does this work for screenwriting too?
Every workflow adapts to screenwriting. The character database becomes a cast bible. The timeline becomes a beat sheet tracker. Continuity checking catches prop, set, and wardrobe inconsistencies. The relationship map tracks character dynamics across episodes.
How many characters can these workflows handle?
Tested with up to 80 named characters across a 120,000-word manuscript. For larger casts, run the Character Database workflow in batches by character tier: main cast first, then supporting cast, then minor characters. Merge the outputs.
What if I don’t have a finished manuscript yet?
These workflows work on whatever you have: outlines, partial drafts, character sketches, worldbuilding notes. The system extracts from whatever sources exist. In fact, running the Story Bible workflow (WF 9) early in your process can reveal gaps in your planning before you write yourself into a corner.
Explore More Categories