All projects
BeginnerFoundations~4h
AI Chatbot
Your first production LLM application
Business problem
Create an AI chatbot that can hold a coherent conversation, maintain history, stream responses, and refuse harmful requests.
Architecture
UI → API (streaming) → LLM provider, with in-memory conversation store and input guard.
Milestones
Track your progress through the project.
0 of 5 milestones complete
Acceptance criteria
- Streaming works — first token < 500ms
- History maintains context across turns
- Persona changes behaviour visibly
- Oversized input is rejected
- Cost is logged per conversation
Deployment checklist
- API key in vault, not env
- Rate limiting per IP
- Streaming with reconnection
- Error logging
- Cost dashboard
Evaluation criteria
- Coherence across 5+ turns
- Latency budget met
- Cost budget met
- Guardrail blocks injection attempts