Abstract
AI Story Generator Novel Maker turns prompts into AI-written stories, but its reviews describe two specific, fixable failures: work disappearing after hours of character-building, and the AI forgetting established story details from one chapter to the next.
Background
With over 1.28 million installs, AI Story Generator Novel Maker lets users describe characters and a plot and generates a story across genres. Reviewers repeatedly describe losing hours of setup work when the app crashes, rotates, or simply fails to save, and the generated story itself losing track of characters mid-way through.
What it does
Generates AI-written stories across genres from user-supplied prompts and characters, with no login required to start.
What changed
Nothing changed recently — this is a persistent reliability gap: unsaved progress lost on crash/rotation, and no long-term memory of characters across chapters.
Why it is soft
The core generation capability works for many users — the complaints are about losing work and continuity, both solvable engineering problems, not a broken core idea.
Strengths
When the app doesn't lose their work, users describe genuinely useful output — decent dialogue, consistent tone within a session, and a fast way to turn a vague idea into a readable draft.
Fast draft generation
"Turned my little prompt to a great story... writes decent dialogue and keeps tone consistent" — the core generation quality is praised when it completes.
No-login option
Several reviewers specifically like being able to start immediately without creating an account, a lower-friction entry point than most AI writing tools.
Market gap
The recurring complaints, most-common first:
Work lost / not saved
The largest cluster by far: hours of character and plot setup vanish on crash, screen rotation, or simply reopening the app, with no draft recovery.
No story continuity
The AI forgets established facts — a character's dog becomes a toddler, people already met are reintroduced as strangers — breaking immersion across chapters.
Forced payment, no real trial
Multiple reviewers describe being charged immediately with no free story ever generated, despite marketing suggesting a trial.
I created my full story twice and This app lost everything. IT IS USELESS.. After another attempt I realised that it is not only reming my data. I believe it attempts to force potential users into a subscription.
The bot can't ever remember the story or how characters are from one chapter to the next. Your character can have a dog but the bot will call it a toddler.
- Autosave every keystroke, locally first. This alone would resolve the single loudest complaint cluster in the reviews.
- Carry a structured character sheet into every generation call. Modern LLM context windows can hold full character/plot state — the continuity failures are a prompting gap, not a model limitation.
- Let people generate one full story before ever asking for payment. Removes the "charged before I saw any output" complaint entirely.
Build complexity
Estimated, not derived: one developer, part-time, no funding and no paid acquisition. The generation itself is a thin LLM API wrapper; the real work is reliability.
| Workstream | Weeks |
|---|---|
| LLM story generation + character-sheet continuity | 1.5 |
| Robust autosave/draft persistence | 1.5 |
| Chapter/character UI | 1.5 |
| Billing (generate-first, pay-later) | 0.5 |
| Total | 5.0 |
Expected revenue
Assumes a $3.99 one-time unlock for unlimited chapters after a fully free first story, given how many reviewers say they'd rather just use a general AI chat app if charged upfront.
| Month 12, monthly | ||
|---|---|---|
| Installs | 4,000 | |
| Activated | 1,800 | 45% |
| Still on at 30d | 700 | 18% |
| Paid | 250 | 6% |
| Revenue | $1,000 |
Biggest challenges
Not the reliability fixes — convincing users this is worth a dedicated app when several reviewers already say they'd just use a general-purpose AI chatbot for free instead.
Competing with free general AI chat
"I'll be better off using Chat-Gpt to create it at this point" is a direct threat — the app needs a mobile-native UX edge, not just a prompt wrapper.
LLM API cost at scale
Long, multi-chapter stories consume real tokens; pricing has to account for genuinely long generations, not just short replies.
AI angle
The product is inherently AI, and the fix to its worst complaint — losing story continuity — is itself an AI-angle: today's larger context windows can hold a full character/plot state across a whole book, which is a markedly cheaper fix than it would have been a couple of years ago.
Without AI
Manual novel writing, or hiring a ghostwriter — the exact slow, expensive process this category exists to shortcut.
With AI
A structured character/plot summary re-fed into a long-context model on every generation call keeps continuity without a custom memory system.
Conclusion
Build it. Two specific, well-understood engineering fixes — real autosave and real continuity — directly answer the loudest complaints, though free general AI chat is real competition.
5 weeks
Time to v1
$3.99
One-time, free first story
$1,000
Revenue at month 12
1.28M
Installs proving demand despite the bugs