Agents
Agents: the model-driven loop, when to choose one over a workflow, guardrails, precise terms and the trade-off.
5 notes · 10 min total
- 1What an Agent IsAn agent is a tool calling loop where the model decides the next step and when to stop. The one structural difference from a workflow.2 min
- 2Workflow or AgentThe rule of thumb: if you can name every step upfront, write a workflow. The two signals that an agent is really needed.2 min
- 3GuardrailsThe three guardrails every agent needs, always together: a maximum number of steps, a budget cap, and human approval for risky actions.2 min
- 4Precise TermsWhat GenAI, agent and agentic each mean, what is not an agent, and how to describe an agentic system precisely in an interview.2 min
- 5The Trade-offAgents handle open-ended tasks but cost more, run slower and are harder to debug than workflows. Why to start with a workflow.2 min