Competitive Landscape — Story / Universe / Character Software

Type: Living document · re-analyzed weekly (space is moving fast) Owner: product (Sandon) Last updated: 2026-07-01 Next review due: 2026-07-08 Update protocol: see §7 Weekly re-analysis protocol. The recurring reminder lives at .claude/scheduled-prompts/2026-07-08-competitive-story-software-reanalysis.md and self-perpetuates each week.

How to read this doc. §1 is the one-paragraph strategic read. §2 is the feature scorecard. §3 profiles each competitor (with a last-checked date per entry). §4 is where we lead / lag. §5 is the plan (workstreams + ranked bets). §6 is the changelog — every weekly pass appends a dated delta. §7 is the protocol for running the weekly update.

1. Strategic read (TL;DR)

Our system is not the same category as the tools it gets compared to. The "story software" market splits into worldbuilding wikis (World Anvil, Campfire, Kanka), AI drafting + story-bible tools (Novelcrafter, Sudowrite, Novarrium, Inkfluence, EPOS-AI), manuscript/organization (Scrivener, Plottr, Dabble), and prose→screenplay (LivingWriter, Squibler). Every one of them stops at "help the writer produce a manuscript."

We are the only stack that treats canon as a production input feeding multi-format output — prose → audiobook (multi-voice) → screenplay (Fountain/FDX) → video (Scene Studio) → print — from one source of truth (the Production IR). That is our moat.

Our two exposed flanks:

1. Writer-facing drafting UX — competitors have spent years polishing the author's editor; ours has been operator/admin-facing. This is now an explicit priority (see the Writing Experience workstream in §5).
2. A handful of "read-layer" bible features (progression timelines, mention tracking, visual relationship webs) that we have the data for but not the UI.

2. Feature scorecard

Legend: ✅✅ leader / differentiated · ✅ solid parity · ⚠️ partial or data-only (no UI) · ❌ absent · ❓ unknown

3. Competitor profiles

Novelcrafter — the story-bible benchmark · last-checked 2026-07-01

• What it is: AI drafting workspace built around the Codex, a wiki-like structured story bible that auto-injects character/location/lore into prompts.
• Flagship strengths: Codex depth ("wiki on steroids"), Progressions (how characters/relationships/politics change over the arc), automatic mention tracking, relations between entries, series-shared Codex, BYOK model choice.
• Weaknesses (from reviews): "powerful but a time sink"; heavy manual entry; steeper learning curve; BYOK means managing separate API costs/config.
• Pricing: $4 (Scribe) / $8 (Hobbyist) / $14 (Artisan) / $20 (Specialist) per month + your own AI API costs (OpenRouter).
• Why they matter to us: closest analog to our Story Bible + Universes. We match/exceed on shared canon (we add per-character images) and lead on enforcement; we lag on Progressions UI, mention tracking, and — critically — the writer-facing editor.

Sudowrite — the prose-quality leader · last-checked 2026-07-01

• What it is: AI writing partner; Story Bible has 7 sections (Braindump, Genre, Style, Synopsis, Characters, Worldbuilding, Outline).
• Flagship strengths: Muse model = best sentence-level creative prose in the category; polished inline tools (Rewrite, Describe, Expand); Smart Character Import (paste ≤60k words); up to 2,000 characters; ≤30 worldbuilding elements auto-imported at a time; series folders share canon.
• Weaknesses: context limits bite on full novels ("60k words in, ask about a chapter-3 character and it struggles"); 2025–2026 pivot is "help me finish," implicitly conceding the long-manuscript memory problem.
• Pricing: annual $10 / $22 / $44 per month (225k / 1M / 2M credits); monthly $19 / $29 / $59. AI bundled.
• Why they matter: the bar for writing experience and prose quality — the thing the user now wants us to beat. Their weakness (long-context consistency) is our architectural strength.

World Anvil — the deep wiki · last-checked 2026-07-01

• What it is: wiki-style worldbuilding platform; dedicated article types for religions, governments, flora/fauna, dozens of categories.
• Strengths: deepest taxonomy + presentation; loved by TTRPG GMs.
• Weaknesses: heavy/complex for straight prose writers; storage-only (no generation enforcement).
• Pricing: Grandmaster ~$12/mo, $105/yr, $650 lifetime.
• Why they matter: depth benchmark — but their depth serves GMs, not an audio/video pipeline, so not worth chasing head-on.

Campfire (Campfire Write) — the novelist-friendly modular tool · last-checked 2026-07-01

• Strengths: modular (buy individual modules), visual relationship webs, good character-arc + manuscript tools, novelist-leaning.
• Weaknesses: fewer specialized templates than World Anvil; storage-only.
• Pricing: all modules $12.50/mo, $125/yr, $375 lifetime.
• Why they matter: relationship-web UX is the feature writers screenshot; we have the data, not the graph.

Novarrium — the consistency-first newcomer · last-checked 2026-07-01

• What it is: AI novel platform betting the company on automatic fact enforcement (9 verification checks/chapter), OCEAN personality modeling, story bible that tracks every detail.
• Why they matter: the only competitor architected like us (canon-as-constraint). Direct threat on our strongest claim. Watch closely.

Also-track (lighter profiles)

• Inkfluence AI — "Smart Continue" includes recent-chapter context while drafting (continuity during, not after). Positions as best continuity checker 2026.
• EPOS-AI — persistent manuscript memory (structured project DB, not a single context window); catches consistency errors across 100k words.
• LivingWriter — AI "Convert to Screenplay" (chapter→screenplay stored beneath original). Closest to our reverse-screenplay engine but blind-transcription, not metadata-grounded.
• Squibler — rapid prose→script, all-in-one, screenplay mode with standard formatting.
• Scrivener — best-in-class organization (binder), multi-POV, series continuity; no AI.
• Plottr — planning/timeline only; exports to Scrivener/Word.
• Dabble — novelist-leaning; plot grid; good for adaptation/treatment/transmedia.
• Highland 2 — Fountain-native screenwriting; plain-text focus.

4. Where we lead / where we lag

Lead (defend + market as a category)

1. Canon-as-constraint, not canon-as-storage — the category's #1 pain ("why your AI novel falls apart after chapter 5"). Only Novarrium matches the architecture; we add the whole downstream pipeline.
2. Multi-format production pipeline — multi-voice audiobook, metadata-grounded reverse screenplay, Scene Studio video, unified Production IR. No competitor renders one canon to five formats.
3. Universes — reusable canon with visual character/place libraries + version history + image identity anchors. Exceeds Novelcrafter's series-shared Codex.
4. Odessa — personalized story from real journal/mood/life-map data. Category of one.

Lag (close these)

1. Writer-facing drafting editor — now an explicit priority. See §5 Workstream A.
2. Progression / character-state timeline — ✅ shipped (B2, 📈 Progression tab, PR pending).
3. Automatic mention tracking — ✅ shipped (B3, 📍 Mentions tab, PR #961).
4. Visual relationship webs — ✅ shipped (B4, 🔗 Relationships tab, PR pending). (Bible-scoped; ai_persona_connections could feed a persona-wide view later.)
5. Model choice / BYOK — admin-configured; needed now that D3 = external creators.
6. JQ Canon Keeper — ✅ shipped Phases 1–3 (canon-mode.ts; the design doc was stale). Our best answer to the "Codex is a time sink" complaint, and now a clear lead (no competitor has conversational canon management). Remaining polish: prose-vs-bible signals in canon_health, universe-level edits, settings/theme write tools.

5. The plan

Four workstreams, ranked. Each bet notes leverage and rough size. Tasks live in the product_tasks DB (mirror at /dashboard/admin/tasks); IDs backfilled here as they're created.

Workstream A — Writing Experience (author-facing editor) — TOP PRIORITY

Goal: match Sudowrite/Novelcrafter on the feel of writing, then beat them on consistency (our architectural edge). The user's explicit ask: "I want the writing experience to be the best as well."

Foundational constraint (user directive, 2026-07-01): build off the existing HiveJournal writing experience as much as possible. Do NOT start a greenfield editor. Reuse the shipped journal/entry composer, the write.cafe plugin embedded in it, the notebook editors, and the rehearsal-room/writers surface as the substrate — extend them toward manuscript drafting rather than replacing them. A1 is "evolve the HiveJournal writing surface into a bible-aware manuscript editor," not "new app." First build step is therefore an inventory of the current writing surfaces + a reuse/extension map before any new UI.

Audience (D3, resolved): external creators → the editor must be polished/self-serve, and the stack should head toward BYOK / model choice + per-user ownership.

• A1. Author-facing drafting editor — a real manuscript editor (not /dashboard/admin), distraction-free, with the Story Bible at hand (reference mid-scene: names, relationships, setting constraints). This is the gate for any external-creator GTM (Emberkiln). Large — needs its own design doc.
• A2. Inline AI actions — Rewrite / Describe / Expand / Continue at the cursor, grounded in the bible. We have the critique loop; this is the inline, low-latency counterpart. Medium.
• A3. Chat-with-your-manuscript in the editor — scope JQ to the open manuscript + bible so "what did Evelyn know in chapter 3?" works inline. Builds on JQ's function-calling. Medium.
• A4. Prose-quality bar — decide our answer to Muse: keep the critique-loop (iterate-and-refine) strategy but benchmark sentence-level output against Sudowrite and close the gap where it's real. Investigation → decision.

Workstream B — Canon intelligence (defend the moat's front door)

• B1. JQ Canon Keeper — ✅ SHIPPED Phases 1–3 (discovered 2026-07-02 already built in canon-mode.ts; the design doc + this plan were stale). Read context + canon_health audit + versioned applyDelta edit tools + cross-story canon_consistency + universe audit, wired into chat-service + Chatbot. Prose-signal folding into canon_health ✅ (unmentioned / drift / orphan refs). Creator unlock ✅ (2026-07-03): canon mode is now own-or-super (not super-admin-only) + grounds in the open chapter's prose (canon_episode_id, migration 447). Settings/theme/voice write tools ✅ (2026-07-03): canon_add_setting / canon_add_theme / canon_add_voice_rule / canon_add_tone_rule / canon_add_avoid (append-only, deduped; versioned via applyDelta). Remaining polish (optional): universe-level edits; update/remove (vs add) for settings/themes/rules. See JQ_CANON_KEEPER.md.
• B2. Progression timeline UI — visualize character state at chapter N over existing data. Read-layer, cheap.
• B3. Mention index — cross-manuscript "every scene X appears in." Read-layer, cheap.
• B4. Visual relationship graph — render relationships[] / ai_persona_connections. Read-layer, demos well.

Workstream C — Production moat (extend the lead)

• Keep pushing the pipeline nobody else has (Scene Studio → short film, Emberkiln render kit). Tracked in SHORT_FILM_STUDIO.md, EMBERKILN_PIPELINE_ARCHITECTURE.md. No new competitive pressure here — it's greenfield vs. the market.

Workstream D — Positioning & GTM

• D1. Market the moat as a category, not a feature list — "Canon → audiobook + screenplay + video from one source of truth" has zero competitors. Don't sell against Novelcrafter on bible depth.
• D2. Price against production value — "cost to produce an audiobook/screenplay," not "monthly cost to write." Higher willingness-to-pay anchor than the $4–59/mo drafting tools.
• D3. Audience decision — ✅ RESOLVED 2026-07-01: external creators. The writing experience targets public authors drafting in our editor (the Emberkiln external-creator GTM). Consequence: A1 must be polished, self-serve, and distraction-free (not a lean internal tool), and the whole stack must eventually support BYOK / model choice and per-user ownership. Size A1 accordingly.

Sequencing recommendation

1. Now: B1 (JQ Canon Keeper) + B2/B3/B4 (cheap read-layer wins) in parallel — high ROI, mostly UI over existing data.
2. Next (gated on D3): A1 editor design doc → build; A2/A3 ride on it.
3. Continuous: C (production moat), weekly competitive refresh (§7).

6. Changelog

Each weekly pass appends a dated entry: what changed in the market, what we updated here, and any new tasks spawned.

2026-07-01 — initial analysis

• Created the doc from the first full competitive sweep (Novelcrafter, Sudowrite, World Anvil, Campfire, Novarrium, + also-track set).
• Established the "different category / production pipeline" strategic frame.
• User set writing experience as an explicit top priority → added Workstream A.
• Set up the weekly re-analysis reminder.
• Decisions: D3 audience resolved → external creators (A1 sized for polished self-serve). Writing experience (Workstream A) must build off the existing HiveJournal writing surface (journal/entry composer + write.cafe plugin + notebook editors + rehearsal-room/writers), not a greenfield editor. Execution kicked off starting with the read-layer wins (B3 mention index first, then B2 progression timeline, B4 relationship graph — "ship one per PR").
• Shipped: B3 cross-manuscript mention index — pure core (apps/backend/src/services/character-mention-index.ts, 14-test golden suite) + GET /api/story-seasons/:id/mention-index + a "📍 Mentions" tab in the Story Bible modal (per-chapter mention counts w/ snippet + canon-health flags: unmentioned characters, first-appearance drift). (PR #961, merged.)
• Shipped: B2 character progression timeline — pure core (apps/backend/src/services/character-progression.ts, 10-test golden suite) + GET /api/story-seasons/:id/progression + a "📈 Progression" tab (diffs each character's status/state/arc/first-chapter across the versioned bible history; introductions + change entries per version). (PR #962, merged.)
• Shipped: B4 visual relationship web — pure core (apps/backend/src/services/character-relationship-graph.ts, 9-test golden suite) + GET /api/story-seasons/:id/relationship-graph + a "🔗 Relationships" tab: dependency-free SVG circular node-link graph (mutual vs one-sided edges, node size = degree, click-to-focus) + orphan-reference canon-health flags. All three read-layer wins (B2/B3/B4) done.
• Discovered already shipped: B1 JQ Canon Keeper (Phases 1–3) — canon-mode.ts + chat-service + Chatbot, migration 417. The design doc + this plan had it as "design only"; corrected both. Scorecard: conversational canon management flipped ⚠️→✅✅ (a lead — no competitor has it). Remaining polish tracked in §5 B1.
• Shipped: prose-eval automation — a model-change-gated cron (prose_quality_eval) re-runs the benchmark as new models come out (fingerprints the model registry; $0 skip unless a model changed or >90d), persists scorecards to prose_eval_runs (migration 446), and flags via a partial heartbeat when a new model beats the current writer tier by >1.0 (the signal to revisit A4-R4). Off by default (spends LLM). So the writing experience's prose quality stays measured over time, not a one-off.
• Shipped: A4-R4 — draft-model tier decided from data. The R3 eval scored Opus 8.55 / Sonnet 8.31 / Haiku 6.47 (default draft trails 2.08, well past the bar). Sonnet = sweet spot (within 0.24 of Opus at ~1/5 cost). The editor's inline prose-assist now defaults to Sonnet (WRITER_QUALITY_MODEL_KEY); the audio pipeline keeps Haiku (cost-first) + Opus on promote. The writing experience's prose quality is now measured + tuned, not assumed.
• Shipped: A4-R3 — prose-quality eval harness. Pure core (prose-eval.ts: 7-dim rubric + blind-judge prompt + score parse/aggregate, 12 golden tests) + on-demand runner (scripts/prose-quality-eval.ts) that ranks draft-model configs with a blind LLM-judge (config hidden, order rotated, ≥3 averaged) and prints a scorecard + the gap-vs-default line. Turns the A4-R4 draft-model-tier decision into a number instead of a vibe.
• Shipped: A4-R2 — sentence_rhythm critic ("Rhythm Reader") in the critic loop, on by default: flags flat cadence (uniform length, no fragments, metronomic clauses) + suggests re-cadenced rewrites. Targets the exact craft dimension 2026 evals say separates polished from machine prose — Muse's headline strength — inside our process bet. Backend-only, auto-surfaces in the critic UI.
• Shipped: A4-R1 — Style Examples (voice exemplar), our analog to Muse's most-loved feature. story_seasons.style_example (migration 445) + a "🎙 Voice" editor in the prose editor + injection into chapter generation AND inline prose-assist ("match this rhythm + diction, not the content"). Closes the biggest felt gap vs Muse.
• Investigated: A4 — prose quality vs Muse (PROSE_QUALITY_BENCHMARK.md). Muse and our engine are opposite bets (fiction-tuned model vs. process/grounding). Decision: don't rewrite the engine to out-Muse one paragraph; ship Style Examples + a rhythm critic, build an eval harness to pick the draft-model tier from data, keep the consistency/pipeline moat. R1–R5 filed as tasks. Prose-quality row stays ✅ for us (different strategy: critique-loop, not best-first-draft).
• Shipped: A3 — chat-with-your-manuscript: a "💬 Ask JQ" button in the prose editor opens JQ in canon mode scoped to the story (above the editor, so you chat while writing); a selected passage is seeded into the chat so you can ask about the actual text. Reuses the existing jq:open-canon plumbing + canon mode — no backend/migration. Scorecard "chat-with-your-manuscript" ⚠️→✅ (in-editor).
• Shipped: A1a-auth — per-season prose-edit ownership enforced on the editor endpoints (season-access.ts canEditSeasonProse, strict own-or-super, 5 golden tests; assertSeasonProseAccess guard on GET/PATCH prose + prose-assist). Closes the "any creator can edit any season's prose" hole. No migration needed — owner_user_id already exists and studio uploads already set it. RLS rewrite deferred (dormant, not the enforcement layer). Unblocks external creators editing their own manuscripts.
• Shipped: A2 — selection-anchored inline AI actions in the prose editor (InlineAiToolbar.tsx + POST /api/story-seasons/:seasonId/prose-assist): select a passage → Rewrite / Expand / Describe / Continue, every action bible-grounded (buildBiblePromptBlock). The Sudowrite table-stakes feature — but grounded in canon we generate from. Scorecard "inline AI actions" flipped ⚠️→✅.
• Shipped: A1b — bible side panel docked in the prose editor (BibleReferencePanel.tsx): scannable Characters/Settings reference (physical, current state, voice tics, relationships) + filter + click-a-name-to-insert-at-cursor. The "codex at hand" feature — now bible reference is live while writing, not in a separate modal.
• Shipped: A1a first slice — direct chapter-prose editing with a TipTap editor (ProseEditorModal.tsx, opened from ChapterTakesModal) + GET/PATCH /api/story-seasons/episodes/:id/prose (versioned manual_edit, best-effort audio rebuild). Fills the "prose is view-only" gap. TipTap chosen as the editor substrate (D-B); serializes to plain text so the pipeline is unaffected. Gated to current creator/super-admin; external-creator route + ownership/RLS spec'd as A1a-auth. Design: WRITING_EXPERIENCE_A1.md.
• Shipped: B1 polish — canon_health now folds in the prose-vs-bible signals from the read-layer cores (unmentioned characters, first-appearance drift, orphan relationship refs) via proseCanonFindings + mergeCanonHealth in canon-mode.ts (13 golden tests). JQ's "any loose ends?" audit is now prose-grounded, not bible-structure-only — connecting B2/B3/B4 into the canon keeper.

7. Weekly re-analysis protocol

Run this every week (reminder self-perpetuates via .claude/scheduled-prompts/). Goal: keep §2–§4 current and append a §6 changelog entry — fast, not a from-scratch rewrite.

Steps:

1. Refresh the incumbents — web-search each of {Novelcrafter, Sudowrite, World Anvil, Campfire, Novarrium} for <name> new features / changelog / pricing in the last ~2 weeks. Update that competitor's profile + bump its last-checked date.
2. Scan for new entrants — search best AI story writing / worldbuilding software 2026, AI novel consistency checker 2026, novel to screenplay AI 2026. Any tool not yet profiled → add a light "also-track" entry.
3. Re-score the matrix (§2) — only cells that changed. If a competitor closed one of our leads (esp. Novarrium on consistency, anyone shipping a production/audio/video feature), flag it loudly in the changelog.
4. Update lead/lag (§4) and, if the plan should shift, note it in §5.
5. Append a §6 changelog entry dated today: market deltas, doc edits, any new/closed tasks.
6. Bump the doc header Last updated + Next review due (+7 days).
7. Commit on a branch + PR (main is protected). Reschedule: copy the scheduled-prompt to next week's date with status: pending, mark this week's fired.

Watch list (things that would change our strategy):

• Any competitor shipping multi-voice audiobook, prose→screenplay, or story→video — that's an attack on the moat.
• Novarrium maturing its consistency guard into a full pipeline.
• A drafting tool (Sudowrite/Novelcrafter) adding a genuinely better long-context consistency story.
• Pricing moves that reset the willingness-to-pay anchor.

Sources (2026-07-01 sweep)

• Novelcrafter Codex · Novelcrafter review · Novelcrafter pricing
• Sudowrite Story Bible docs · Sudowrite Characters docs · Sudowrite pricing
• Campfire vs World Anvil (Kindlepreneur)
• Best AI fiction tools 2026 (MyLifeNote)
• Best AI novel continuity checking 2026 (Inkfluence) · Novarrium — AI story consistency
• LivingWriter chapter→screenplay </content>