Can I vibe code ChatBot?

Build a custom lightweight AI customer support chat system for an e-commerce store. 1. Architecture: Create a Node.js/TypeScript backend API (Fastify) and an embeddable plain JavaScript widget wrapper. Store conversation history, embeddings, and chat sessions in PostgreSQL with pgvector. 2. Data Vectorization: Build a script to parse store policy documents (Markdown/PDF) and product catalog data. Generate vector embeddings using OpenAI text-embedding-3-small and store them with metadata tags in pgvector. 3. Chat Agent Logic: Implement an endpoint `/api/chat` using OpenAI GPT-4o with function/tool calling. Define tool functions: `get_order_status(order_id, email)`, `get_return_policy()`, and `search_products(query)`. Execute similarity search on pgvector when answering general FAQ questions. 4. Customer Authentication: When a user requests order status, require both order number and matching customer email before making API calls to the e-commerce platform. 5. Fallback & Escalation: If the model returns low confidence or the user requests a human, mark the session status as 'escalated', capture customer email, and trigger a Webhook to send an email notification or Slack message. 6. Admin Interface: Build a simple React dashboard displaying chat transcripts, escalation queues, and token usage analytics. Out of scope: Visual drag-and-drop conversational node designer, multi-tenant workspace management, and native mobile apps.

$ each button prefixes agent-specific run instructions · build your own product, never copy proprietary code, trademarks or designs