Skip to content
NTPU AI Learn

AI 202 Apply

Design how a real agent plans, reasons and adapts, with a Canvas

What the Canvas is for

The Agent Planning & Reasoning Canvas is the capstone of AI 202. For one real use case you define the goal and state, decompose it into subgoals, choose the actions and the planning strategy, decide how results are evaluated and when the agent reflects, name the re-planning triggers, and set the limits, escalation and stop conditions. Human, AI, Human: a person defines the task, the agent assists or executes, and a person reviews and stays responsible.

What you fill in

  • Agent Goal, Initial State and Observations
  • Subgoals and Available Actions, with irreversible or high-impact actions marked
  • Tools (connects to AI 201, without repeating the MCP Tool Canvas)
  • Planning Strategy and Decision Policy
  • Evaluation and Reflection
  • Re-planning Triggers
  • Constraints & Boundaries
  • Human Escalation and Stop Conditions
  • Final Output, and the reflection question

Warning

An irreversible or high-impact action should never run without a person's approval. State which outputs and actions a person must confirm before anyone relies on them.

Examples

The canvas page shows a worked example (choosing a workshop venue) and ideas for other occupations: customer support, ESG report review, research literature scan, financial analysis and administrative operations. All use simulated data.

Before you submit

Warning

Do not enter secrets, credentials, personal data, client data or sensitive information you are not authorized to share. Use simulated or public examples only.

What comes next: AI 301 and AI 302

AI 202 teaches how one agent plans, reasons, evaluates and adapts. Later courses in the curriculum, AI 301 (Multi-Agent Collaboration & Orchestration) and AI 302 (Agent Harness Engineering & Reliable Automation), build on it. See the course catalog for all courses.

Academic mapping: M5276 Week 10

AI 202 maps primarily to Week 10 of M5276 Agentic AI Innovative Applications: Agentic AI Planning & Reasoning Engines: ReAct, Search, MCTS, Self-Reflection and Adaptive Planning. The topics covered are planning, ReAct, search, the MCTS concept, self-reflection, adaptive planning and re-planning.

Multi-agent architecture remains primarily in Week 13, so it is not part of this course.

Go to the canvas

What you need to complete

  • Complete all seven AI 202 modules
  • Score at least 80% on the course post-test
  • To earn the Planning & Reasoning Capability Certificate, also submit an Agent Planning & Reasoning Canvas and have it reviewed and passed by a Credential Admin against the rubric

How you are assessed

A Credential Admin scores the Agent Planning & Reasoning Canvas against eight weighted domains: Goal & Subgoal Framing, State & Observation Design, Action & Tool Design, Planning Strategy & Decision Policy, Evaluation & Reflection, Re-planning & Constraints, Human Escalation & Stop Conditions, and Final Output & Key Reflection. A score of at least 19 out of 24, with no critical-fail rule triggered, is a PASS; 16 or below, or any triggered critical-fail rule, is a FAIL.

Info

This human-reviewed rubric is different from the instant, automated feedback score you see right after submitting: that instant score is for your own reference, while the Capability Certificate is only ever issued after a Credential Admin reviews the canvas against this rubric.

Info

This scoring logic is published so you know what is being assessed. It is not the same as revealing quiz answers.

Practical evidence

  • A recorded rubric result (PASS, REVISION_REQUIRED, or FAIL) with no critical-fail rule triggered
  • A PASS records capability evidence separately for each of the eight fine-grained AI 202 competencies (Workflow vs. Adaptive Agent Diagnosis, Goal Decomposition & Planning, ReAct: Reasoning + Acting, Search, Trees & MCTS, Evaluation & Verification, Self-Reflection, Adaptive Planning & Re-planning, Agent Planning & Reasoning Canvas Design)

What you can earn

AI 202 Course Completion Certificate: earned once you complete all seven modules and meet the post-test threshold, with no Canvas submission required. Planning & Reasoning Capability Certificate: requires a separately submitted and reviewed Canvas. The two are independent and do not affect each other. The existing AI 202 completion certificate mechanism (/pr/certificate) is unchanged and additionally requires all 6 practice activities and at least one Canvas submission -- a stricter, separately-standing completion check.

What this credential does not mean

This credential confirms the holder completed AI 202's required course and post-test, or (for the certificate) demonstrated goal decomposition, planning strategy, evaluation/reflection, and human-escalation design on the Agent Planning & Reasoning Canvas.

  • It does not mean the holder holds a formal AI agent development or deployment professional license or is qualified for a specific role
  • It does not mean any specific agent the holder designs later is automatically safe or correct
  • It does not mean human review can be skipped before a high-impact or irreversible action (Human, AI, Human)

What next?