I need a conversational AI that feels the same whether a visitor is on my website, chatting through WhatsApp or Telegram, or asking questions from the checkout page of an e-commerce store. The project centers on one unified chatbot engine, with channel-specific wrappers so every platform shares intents, entities, and business logic. Core expectations • Design the dialogue flows, intent recognition, and entity models • Build or configure the NLP layer (Dialogflow, Rasa, Microsoft Bot Framework, Amazon Lex, or an open-source alternative—whichever fits best without painful lock-in) • Expose secure APIs and webhooks so the bot can pull order info, FAQs, and account data in real time • Integrate the bot widget into the website, hook it into major messaging apps, and connect it to my e-commerce storefront • Set up logging, analytics, and a confidence-based hand-off to a live agent Deliverables 1. Source code and configuration files 2. Documentation that shows how to add new intents and retrain models 3. A short hand-over call or video walk-through Acceptance criteria – Median response time under 1 s in production – At least 90 % correct intent detection on a held-out test set – Smooth escalation to human support when confidence falls below 0.4 Preferred languages are Python or Node.js, though I am flexible if another stack suits the brief better. Milestone suggestions and a realistic timeline are welcome.