AI System Design
Designing complete, production-grade AI systems.
- How to Design an AI SystemAdvanced
Design an AI system in order — job and SLO, shape, knowledge and tools, eval, security, traces, then serving and cost — and do not start with a trophy model.
- Design a RAG ChatbotIntermediate
A RAG chatbot retrieves an approved corpus, reranks, and answers with real citations — not an agent, not a fine-tuned wiki.
- Design an Enterprise Knowledge AssistantAdvanced
An enterprise assistant is RAG plus identity — ACL at retrieve, provenance on ingest, citations the user is allowed to open, and no shared bot token.
- Design a Customer-Support AgentAdvanced
Support is identity, policy RAG, a structured action preview, and a human or policy gate — task success is a correctly resolved ticket, not a thumbs-up.
- Design a Coding AgentAdvanced
A coding agent explores with repo tools, patches, and verifies in a sandbox — tests are the judge, the PR is HITL, and the repository is not a paste into the window.
- Design a Research or Multi-Agent SystemAdvanced
Default to one researcher with search tools; add parallel gatherers only when sub-questions are independent — cap cost, keep citations, and don't share god tools.
- Design an AI Search EngineAdvanced
AI search is query, hybrid retrieve, rerank, then grounded snippets — a chatty generator without ranking metrics is not search.