AI Agents
AI agents combine large language models (LLMs), tools, and memory to complete tasks in chat UIs or automated workflows. This guide introduces patterns used alongside cloud platforms and open-source tooling.
Agent loop
- User sends a prompt to the agent
- Agent plans and calls tools (HTTP APIs, databases, MCP servers)
- Tool results return context; the agent produces a response
Core building blocks
| Component | Purpose |
|---|---|
| System prompt | Role, constraints, safety tone |
| Tool routing | Function calling, MCP, allowlists |
| Memory | Session state, user preferences |
| Evals | Regression tests for agent behavior |
Guides in this section
- RAG pipelines — ground answers in your data
- Production AI systems — guardrails, cost, observability
Related topics
- Deployment automation — ship agent services
- SEO — document public agent APIs and docs sites