Part III

Build

Connect the interface, data, business logic, and feedback in one usable slice.

Example: an agent retrieves the right policy, checks eligibility, and proposes a resolution for review.

Chapter 5

Build a thin vertical slice

Updated

Deliver one complete user action through the system.

Read

  • Building effective agents Compare fixed workflows with model-directed agents. The article notes that its tooling has evolved.Focus: Compare routing, chaining, orchestrator/worker, evaluator/optimizer, and autonomous agent patterns. Apply: Explain why a deterministic implementation or fixed workflow is insufficient before adding an agent loop.
  • The AI-native interview The employer’s published plan, build, review format and a debugging-interview pilot.
  • Writing effective tools for agents — with agents Tool interfaces are a measurable part of agent performance.Focus: Read tool selection, namespaces, response content, pagination, and held-out evaluation. Apply: Compare two tool contracts on task success, invalid arguments, redundant calls, latency, and token cost.
  • 12-Factor Agents · Unify execution state and business state Understand the state that a resumable workflow must preserve.Focus: Read factor 5, then use the author-linked 12-Factor Agents talk as an alternative overview. Apply: Specify the persisted events needed to resume a paused task and reconcile any uncertain external side effect.

Do

  • Compare deterministic logic, a fixed AI workflow, and an agent for the task.
  • Build the smallest end-to-end slice with a usable interface.
  • Demo it to a user and record the change you made from their feedback.

You are done when: A working path from input to outcome, with assumptions and limitations written down.