Deploy Your Custom AI Chatbot

Build a support assistant for a bicycle shop using an approved knowledge base. Answer questions about repairs and opening hours, cite the relevant article and offer a human handoff when information is missing.

Include these capabilities:
- Chat interface with streaming message display
- Conversation history and session management
- Customizable system prompts and personas
- Embeddable widget for any website
- API integration with OpenAI, Claude, and more

Build the complete working project with its routes, data model and interactions. For saved or private data, implement persistence, authentication and server-side permissions. Keep API credentials on the server. For external services, document the required setup and use test mode until real accounts are connected. Do not present a simulated integration as a completed live connection.

Make the interface responsive, with accessible labels, keyboard navigation and visible focus. Include loading, empty, validation and error states. Use clearly labelled sample data, never invented customer testimonials or results. Choose a visual direction suited to this project and keep typography, spacing and controls consistent. Explain how to run, test, export and deploy the source.

Verify this end-to-end flow:
Ask a covered question and an unsupported one. Confirm the cited source exists, missing information is acknowledged and the handoff contains the conversation.