Skip to content
← Projects

Listify

A multimodal AI draft workspace that turns one product photo and optional seller context into editable Indonesian marketplace listing copy.

Open live app View source

Listify interface preview enlarged

Listify is a focused AI draft workspace for Indonesian marketplace sellers. It turns one product photo plus optional seller context into editable listing copy for marketplaces such as Shopee, Tokopedia, TikTok Shop, or similar platforms.

The product is intentionally small: it helps sellers get a useful first draft without learning prompt engineering, managing a catalog system, or connecting directly to marketplace APIs. Sellers review, edit, and copy the result manually.

The problem

Early-stage UMKM sellers often need clear titles, descriptions, selling points, SEO keywords, category suggestions, and pricing guidance, but the product information usually starts from a photo and a few notes. Asking an AI assistant directly can work, but the result may invent details, miss marketplace structure, or require repeated prompt tuning.

Listify narrows that workflow into a single draft session. The app asks for one product photo, accepts optional context, and returns a structured listing draft that is easy to edit before use.

Product flow

  1. Upload one product photo.
  2. Add optional seller context such as product name, condition, marketplace preference, and notes.
  3. Generate a listing draft through the server API route.
  4. Review and edit the title, description, selling points, SEO keywords, category suggestion, and price guidance.
  5. Copy marketplace-ready text or a fuller seller summary for manual use elsewhere.

What I built

  • Multimodal generation flow from product photo to structured Indonesian listing copy.
  • Server-side request and response validation with Zod.
  • Gemini-powered listing generation through @google/genai.
  • Editable review interface so sellers can correct the draft before copying it.
  • Honest uncertainty handling so the app does not present guessed product details as facts.
  • Lightweight no-storage architecture for a focused MVP.
  • Manual copy-out flow instead of marketplace publishing or catalog management.

Key technical decisions

The app uses Next.js App Router with a focused draft flow on the client and a server API route for AI generation. The browser converts the uploaded product image into base64, combines it with optional seller context, and sends the request to /api/generate-listing.

The API route validates the request, builds the prompt, calls Gemini Flash, validates the structured response, and returns editable draft data. Listify does not persist photos, drafts, or draft sessions, which keeps the MVP lightweight and avoids storing seller product images.

The most important product constraint is trust. Listify is a draft assistant, not an autopublishing tool or market research engine. It should surface useful copy and uncertainty, then let the seller decide what is accurate enough to use.

Current status

Listify is active and deployed. The current version supports one temporary draft session and manual copy-out. It does not include marketplace publishing, saved listing history, database storage, product photo editing, or live market research.