Multimodal chat assistant with GPT-4o for text, images, and PDFs

Chat with thing

This n8n template lets you build a smart AI chat assistant that can handle text, images, and PDFs — using OpenAI's GPT-4o multimodal model. It supports dynamic conversations and file analysis, making it great for AI-driven support bots, personal assistants, or embedded chat widgets.

---

🔍 How it Works

- The chat trigger node kicks off a session using n8n's hosted chat UI. - Users can send text or upload images or PDFs — the workflow checks if a file was included. - If an image is uploaded, the file is converted to base64 and analyzed using GPT-4o's vision capabilities. - GPT-4o generates a natural language description of the image and responds to the user's question in context. - A memory buffer keeps track of the conversation thread, so follow-up questions are handled intelligently. - OpenAI’s chat model handles both text-only and mixed media input seamlessly.

---

🧪 How to Use

- You can embed this in a website or use it with your own webhook/chat interface. - The logic is modular — just swap out the chatTrigger node for another input (e.g. form or API). - To use with documents, you can modify the logic to pass PDF content to GPT-4 directly. - You can extend it with action nodes, e.g. saving results to Notion, Airtable, or sending replies via email or Slack.