5 min read
Agentic AI Chapter 2: The Agent Loop
How an agent actually works: the LLM decides, the runtime validates and executes, and the loop repeats until the goal is met. Walked through turn by turn with the iPhone 18 launch-day example.
Topics
#study-notes
Things I've written about #study-notes
- ai(6)
- gen-ai(5)
- product(5)
- agent(4)
- agentic(4)
- ai-agent(4)
- study-notes(3)
- feature(3)
- kpi(2)
- access(1)
- control(1)
- rbac(1)
- rubac(1)
- abac(1)
- iam(1)
- identity(1)
- dac(1)
- pip(1)
- pdp(1)
- pep(1)
- artificial(1)
- intelligence(1)
- incident(1)
- sla(1)
- l2(1)
- l3(1)
- escalation(1)
- operation(1)
- metrics(1)
- sass(1)
- mrr(1)
- arr(1)
- clv(1)
- cac(1)
- bdfu(1)
- nps(1)
- customer(1)
- saas(1)
- system-design(1)
- architecture(1)
- backend(1)
- database(1)
- security(1)
- prioritize(1)
- agile(1)
- scrum(1)
- backlog(1)
- refinement(1)
- team(1)
- rag(1)
- llm(1)
- beginner(1)
- learning(1)
- rca(1)
- diagnosis(1)
- moscow(1)
- value(1)
- rice(1)
- kano(1)
- eisenhower(1)
- cod(1)
- wsjf(1)
4 min readAgentic AI Chapter 3: Tool Calling
The model never runs a tool. It only asks for one in a structured format, and the runtime validates and executes it. Explained with the iPhone 18 launch-day example.
6 min readAgentic AI Chapter 1: Agent vs Workflow
Before building an agent, ask whether you need one. A workflow follows fixed steps. An agent decides what to do next. Explained with an Amazon returns example.