ListaAI

Brazil has over 12 million active marketplace sellers, most of them micro-entrepreneurs and small shop owners who list products manually. Writing an effective marketplace title and description isn't trivial: it requires knowing each platform's search algorithms, using the right keywords, and structuring information in a format that converts. The reality for most sellers: generic descriptions copied from suppliers, poorly optimized titles, and hours wasted every week on product content. The cost of a bad description is direct: less visibility, less conversion, fewer sales.

Demand for marketplace copywriting tools in Brazil is high and underserved. Generic tools like ChatGPT require sellers to craft prompts, interpret outputs, and adapt tone, creating a barrier for those unfamiliar with AI. Mercado Livre-specific solutions are scarce, and when they exist, they focus on pricing or inventory, not content. The ideal entry point: a 3-field flow (name, specs, platform) → result in under 10 seconds.

• Sellers lose 30 to 60 minutes per day just writing product descriptions
• Poorly written titles are the main cause of low organic ranking in marketplaces
• Most sellers don't know which keywords to use and don't have time to research
• ChatGPT is already used by some sellers, but the free-form prompt friction is a real barrier
• Mercado Livre and Shopee have distinct rules and formats, so a tool that understands this has a clear advantage

Landing page with the value proposition for Brazilian marketplace sellers

ListaAI was built with a minimalist usage flow: the seller provides the product name, technical specs, target platform (Mercado Livre or Shopee), and preferred tone of voice. The AI generates an optimized title with high-volume keywords and a complete description formatted for the chosen platform. The result can be regenerated with one click, copied directly, and is saved in history. Every detail was designed to minimize friction, so the seller doesn't need to understand AI to use it.

Each feature was prioritized based on one question: does this reduce seller time or increase output quality? Platform-specific generation ensures the output respects Mercado Livre's and Shopee's rules and formats. Voice tone adjusts the description from formal to casual. Built-in SEO, with no extra setup, analyzes specs and picks terms with the best ranking potential. Smart regeneration produces a new result without re-entering data. Full history gives sellers control over everything that's been generated.

• Platform generation, with output adapted to each marketplace's rules
• Configurable tone of voice, including formal, technical, casual, or persuasive
• Built-in SEO, with high-volume keywords automatically embedded in the title
• Smart regeneration, providing a new result with one click, no need to rewrite data
• Full history, where all descriptions are saved and accessible in the dashboard

The product was entirely built with AI coding, with Claude as pair programmer at every stage, from scaffolding to deployment. Stack: Next.js 14 (App Router) on the frontend, API Routes for the backend, Prisma + PostgreSQL for data persistence and history, Claude API (claude-sonnet) for content generation, and Vercel for continuous deployment. The generation prompt was the most iterated artifact: each version was tested with real products across different categories, including electronics, fashion, and home, until output quality became consistent.

Freemium model: Free R$0 with 5 descriptions/month, PRO R$39/month unlimited

The key learning was about the relationship between interface simplicity and prompt quality. The simpler the form for the user, the more work the prompt needs to do to compensate for missing context. The solution was a layered instruction system: the base prompt defines the AI's general behavior, while dynamic variables inject the specific context for each generation. The freemium model was chosen to maximize adoption, as the product sells itself on the first use.