AI Agents and Multi-Agent Systems · Lesson 9
Entering into Agentic AI
Agentic AI describes systems that pursue goals with autonomy: they plan multi-step work, use tools, adapt to results and run for longer stretches with less human input.
Real examples include coding agents that edit and test repositories, research agents that browse and synthesise, and operations agents that triage tickets.
Key concerns are reliability, observability, permissions, cost and safety. Treat an agent like a new employee: limited access, clear instructions, and review of its work.
Key points
- Autonomy across multi-step goals
- Coding, research and operations agents
- Guard permissions and budgets
- Observe and review agent work