2
4 Comments

10 days in, 0 users: I built a tool that writes image prompts for you through conversation

I launched StoryRole AI 10 days ago. Zero registered users so far. I haven't done any marketing because honestly, I have no idea how. None of my friends work in text-to-image, and my Reddit account is too new for my posts to even show up. So here I am.

What I built and why

I kept running into the same problem with AI image generation: I had a clear picture in my head but couldn't translate it into a good text prompt. I'd spend 20 minutes tweaking a prompt, generate something that looked nothing like what I imagined, tweak again, and eventually settle for "close enough."

So I built StoryRole AI (https://storyrole.ai). Instead of writing prompts, you just have a conversation. You describe your character however you want, rough ideas, vague feelings, whatever. The AI asks you clarifying questions, helps you pin down the details you couldn't articulate, and then generates the text-to-image prompt for you. You never touch a prompt yourself unless you want to.

The part I think is actually interesting

Most image generation tools treat each image as a one-off. StoryRole AI organizes everything into projects with a cast of characters. Every character in the same project shares context, so when you generate a new character, the AI already knows what the rest of your cast looks like. The visual style, proportions, and aesthetic stay consistent across the entire project without you having to repeat yourself.

That project-level consistency is what got me started, and there's a lot more I want to build on top of it. But right now the core workflow is solid: conversation, prompt generation, and a consistent cast across your project.

The tech (for anyone curious)

  • Next.js 15 + FastAPI backend
  • Qwen LLM for the conversation layer
  • Qwen and Wanx for image generation
  • Supabase (PostgreSQL + Auth)
  • Deployed on Vercel + Railway
  • Three languages: English, Simplified Chinese, and Traditional Chinese (with natural Hong Kong Cantonese-style expressions, since I'm based in Hong Kong)

I built this as a solo founder, using AI coding tools while maintaining full control over the architecture and product decisions.

Where I am now

  • Live at https://storyrole.ai with working pricing (4 tiers, 7-day free trial)
  • Everything I claim on the site actually works
  • 0 users, 0 revenue
  • Zero marketing experience. This post is literally one of my first real attempts to put it in front of people
  • Google Search Console shows almost no impressions

What I'm trying to figure out

  1. How do you find your first users when your product serves a niche (creators who do character design / visual storytelling) and you have no existing audience?
  2. For anyone who does text-to-image generation: does the "conversation instead of prompt writing" approach sound useful, or is prompt crafting actually a skill people enjoy and don't want replaced?
  3. Any feedback on the site itself? I'm genuinely open to hearing what doesn't work.

This is not a roleplay platform. I know the name might suggest that. It's a design tool for turning the characters in your head into consistent artwork.

Would really appreciate any thoughts, even if it's "this probably won't work because X."

on April 28, 2026
  1. 1

    I know a few creators who do character design and visual storytelling who might be willing to answer your questions for free about prompt crafting. Happy to pass some questions along if you'd like.

  2. 1

    The product is stronger than the name.

    “StoryRole” sounds like roleplay software, not a serious creative tool.

    That’s probably costing you the first click before the workflow even gets judged.

    What you actually built is a character generation system with memory, consistency, and visual continuity.

    That’s much closer to a product like Auryxa.com or Zenvarya.com than “StoryRole.”

    Cleaner name, better first impression, less ambiguity. That alone likely improves conversion before traffic does.

    1. 1

      Appreciate the honest read - the roleplay association is something I've been thinking about too, since the product is the opposite of roleplay (it's a structured character + project workflow for visual storytellers, with consistency across an entire cast). I'll be sharpening the tagline and landing copy to make that gap obvious within the first second. Renaming this early is on the table long-term, but for now I think the bigger lever is positioning. Thanks for taking the time to look.

      1. 1

        That’s fair short-term.
        But this is usually the tradeoff:
        better copy can reduce ambiguity
        it rarely fixes the first impression the name already created
        So if the roleplay association is already forming before the tagline lands, the name is still doing more damage than the copy can fully recover.
        Worth fixing with positioning now.
        Still worth pressure-testing with names before that gets more expensive later.

Trending on Indie Hackers
The hardest part isn't building anymore User Avatar 88 comments I sold $6,773 in 2 weeks, with almost no existing community. User Avatar 60 comments Before you build another feature, use this workflow User Avatar 40 comments Ferguson is LIVE on ProductHunt today... so I audited their homepage first! User Avatar 38 comments Built a local-first Amazon profit-by-SKU + QuickBooks/Xero journal tool. Looking for founding users. User Avatar 32 comments I spent months chasing clients who already had a webmaster. So I built something that only finds the ones who don't. User Avatar 26 comments