MenuPlan AI - Sell Project

I'm Manuel, the creator of MenuPlan AI. I built it around a household routine: keep the recipes you actually cook, decide the week's lunches and dinners, then prepare the shopping list from that plan. The Android app is live on Google Play and currently free, without advertising or subscriptions. It supports English, Spanish, French, German, Italian and Portuguese. This is a sale of an early-stage product and agreed assets, with a small recorded user base and no revenue to date. What a user can do today 1. Build and maintain a recipe collection. Add recipes and ingredients, search and filter the collection, and keep preparation instructions with a recipe. Import a recipe from a compatible website, or use AI-assisted import from a photo of a written recipe or pasted text, or generate a draft recipe from a dish name. Users review the result before saving. Recipe exports include readable TXT/PDF files and a JSON backup that can be restored; TXT and PDF are not restore formats. 2. Plan lunches and dinners with specific rules. Select dates, keep fixed dishes, mark days or meals to skip, choose quick/normal/elaborate recipe preferences, and set weekly category targets. Build a menu from the saved recipes, change individual dishes and reuse saved planning patterns. The automatic planner is a local rule-based algorithm: it does not ask an LLM to invent the week's menu. 3. Shop from the plan. Generate the shopping list when ready: ingredient quantities scale with servings and compatible units are combined. Organize it by aisle, add or edit items and check off purchases. If the menu changes, the app prompts the user to update the list; it is not continuously regenerated in the background. Household members can share menus, recipes and shopping lists. The pantry feature can exclude selected staples from shopping; it is not a stock system that tracks quantities or expiry dates. 4. Review and reuse the household's data. View menu history, recipe frequency and weekly summaries. Nutrition displays are approximate and depend on the available ingredient data; missing data is not complete nutrition coverage. Guest mode allows local exploration on one device without an account; its statistics tab is hidden. Already-loaded recipes, plans and lists can be used offline. Shared household sync requires an account and connectivity. The accompanying website explains the product and includes editable weekly planners in English and Spanish. Those planners save in the browser and are separate from the Android app; there is no web-to-app synchronization claim. What the AI actually does The user-facing AI tools cover import from written-recipe photos and text, generation of draft recipes from dish names, drafting preparation steps from a recipe's ingredients, and suggesting grocery aisles for individual items or a batch. AI needs sign-in, a verified email and an internet connection; it is not part of guest mode. Outputs can need correction. The current repository routes photos to Anthropic Claude Haiku 4.5. Text tasks use Cloudflare Workers AI with Llama 3.1 8B Instruct Fast and a Claude fallback for technical failures. Authentication, caching and per-user daily quota code are present. These are the configured implementation details, not a promise of unlimited usage or a verified monthly operating cost. Compatible website imports parse structured recipe data such as JSON-LD or microdata. They do not use AI. The weekly planning algorithm and local browsing also do not require an AI request. This separation gives a buyer a concrete starting point for controlling inference costs while keeping the core routine usable without them. Product and technical assets The application uses React, TypeScript, Vite and Capacitor for Android. Firebase handles identity and shared data. A Cloudflare Worker, KV and Workers AI support backend operations, with Anthropic for selected AI requests and Resend for contact email. The marketing site uses React with Vinext/Next.js-compatible routing, Vite and Sites deployment. The proposed package includes the app source and Android build project, backend code and configuration, marketing-site source, screenshot-generation tooling, documentation and store/brand materials whose ownership and transfer rights are confirmed during diligence. There are six real UI screen types and 60 prepared images across ten regional store-asset sets; this means six app languages, not ten different interfaces. Release preparation for Android 1.2.5 (27) records 568 passing app tests and signed artifacts. The latest recorded Play state is review; the new version is not claimed publicly available here. Those checks are evidence of the tested scope, not a guarantee that every device or cloud-service flow is defect-free. Traction and current finances Recorded Play Console figures are 21 cumulative acquisitions as of 25 September 2026, 15 users with the app installed as of 26 September, and 4 active users in the 28-day window ending 24 Septem...