Creating a full AI assistant like DeepSeek is a massive unde...
Prompt
Creating a full AI assistant like DeepSeek is a massive undertaking, but here is a comprehensive prompt you can give to an AI (like me!) to generate a roadmap, code structure, or specific features for your app. Copy and paste the prompt below. It is designed to cover the Frontend, Backend, AI Integration, and Monetization. --- The Master Prompt ``` I want to build a mobile and web application similar to the DeepSeek AI Assistant. Please act as my Lead Software Architect and provide a complete technical blueprint and development roadmap. Specifically, I need you to generate the following: 1. **Tech Stack Recommendations:** - For the Frontend (React Native for mobile & Next.js for web). - For the Backend (Node.js/Python). - For the Database (Vector DB for RAG, and Relational DB for users). - For the AI/ML (Open-source models vs. API access like OpenAI/Anthropic). 2. **Core Features & Architecture:** - User Authentication (Email, Google, Apple). - Chat Interface (Streaming responses, markdown rendering, code highlighting). - File Upload Support (PDF, DOCX, TXT, Images). - Context Management (How to handle long conversations and token limits). - "Deep Search" functionality (Web search integration using APIs like Tavily or SerpAPI). 3. **The "Brain" (RAG Implementation):** - Provide a step-by-step guide on setting up a RAG (Retrieval-Augmented Generation) pipeline so the AI can answer questions based on uploaded documents. 4. **Database Schema:** - Write the Prisma or SQL schema for `Users`, `ChatSessions`, `Messages`, and `Documents`. 5. **Cost Optimization:** - Strategies to reduce API costs (caching, using smaller models for simple queries). 6. **Sample Code:** - Write a simple Node.js/Express or Python/FastAPI endpoint for handling a streaming chat request. - Write the React component for the chat input box that handles "Enter to send" and "Shift+Enter for new line". 7. **Monetization Strategy:** - How to implement a credit system or subscription tier (Freemium) using Stripe. Please format the response with clear headings, bullet points, and code blocks. Focus on scalability and user experience. ``` --- Why this prompt is effective: 1. Structured: It breaks down the massive app into digestible parts (Frontend, Backend, AI Logic). 2. Technical: It uses correct terminology (RAG, Vector DB, Streaming, Tokens) to ensure you get a useful response, not a generic one. 3. Actionable: It asks for actual code snippets (the API endpoint and the UI component), giving you a head start on coding. 4. Business Aware: It asks about costs and monetization, which is crucial for sustainability. If you want a Super Simple MVP (Minimum Viable Product) prompt: If the above is too big, use this shorter one to start: "I want to build a basic AI chat app like DeepSeek. Give me a step-by-step guide to build it using React for the frontend and Node.js for the backend. I want to use the OpenAI API. Provide me with the code to send a prompt and display the streaming response. Also, tell me how to add a simple 'upload PDF' feature." Next Steps: Once the AI gives you the plan, pick Part 2 (Core Features) or Part 6 (Sample Code) and ask the AI to elaborate on that specific section first. Build it piece by piece!
A system prompt was added to support web rendering
