I want a small proof-of-concept built in n8n that shows an OpenAI-powered agent chatting with visitors on a website and remembering the conversation context. The scope is deliberately simple: the bot only needs to answer customer questions through a website chat widget; no order handling or recommendations are required right now. What I need from you: • An n8n workflow that calls the OpenAI API, stores short-term memory (you can use n8n’s built-in data store or any lightweight database you prefer), and returns the reply to the chat front end. • A quick front-end demo page (plain HTML/JS is fine) so the conversation can be tested live. • A short screen-capture video showing the agent in action—asking two or three questions, watching the bot respond, and demonstrating that it remembers earlier context. • A brief README with setup steps, environment variables, and where to paste API keys so I can reproduce the result locally. Keep the workflow clean, well-commented, and organised into logical nodes (trigger, memory fetch-update, OpenAI call, response). If you already have a similar snippet, feel free to build on it—speed is more important than fancy UI polish. I’ll spin it up on my own n8n instance once you send the export file, the video, and the setup notes.