Agentic AI Planning & Reasoning
Move from fixed agentic workflows to agents that plan and adapt on their own: understand ReAct, search, MCTS, self-reflection and adaptive planning, learn to design how an agent chooses its next step from its goal, context, environment state and results, and complete your own Agent Planning & Reasoning Canvas.
Course storyline: how an agent decides what to do next
The whole course follows one line: observe, reason, plan, act, evaluate, reflect, re-plan.
- 1. Observe
What does the agent see right now?
- 2. Reason
What does that information mean?
- 3. Plan
Which paths are possible?
- 4. Act
What should the next action be?
- 5. Evaluate
Did that step work?
- 6. Reflect
Was the action wrong, or was the plan wrong?
- 7. Re-plan
Continue, change the plan, escalate to a human, or stop?
Your next step
AI 202 · Module 1: From Workflow to Adaptive Agent
Curriculum
- Module 1From Workflow to Adaptive AgentWhen is a workflow not enough, and the agent must decide the next step itself?
- Module 2Goal Decomposition & Planning FundamentalsHow does an agent turn a goal into an executable plan?
- Module 3ReAct: Reasoning + ActingWhy can an agent not simply think first and act later, instead of reasoning, acting and observing in turn?
- Module 4Search, Trees & MCTS for Agent DecisionsWhen an agent has several possible actions, how does it search and compare paths?
- Module 5Evaluation, Verification & Self-ReflectionAfter a step, how does an agent judge whether the result can be trusted and whether the strategy must change?
- Module 6Adaptive Planning & Re-planningWhen should an agent stop following the original plan and re-plan?
- Module 7Agent Planning & Reasoning CanvasHow do you design an agent's goal, state, actions, reasoning, evaluation and re-planning into an implementable architecture?
Progress is stored in this browser and does not identify you. It is uploaded only if you join progress sync below.
Sync my progressJoin a session to keep progress across devices and share it with your instructor
Practice and assessment
- Activity 1: Workflow or Adaptive Agent?Judge which situations a fixed workflow can handle and which need an agent to choose the next step from the current state.Not done yet
- Activity 2: Goal → Subgoal → ActionBreak a real goal into subgoals, actions, constraints and success criteria.Not done yet
- Activity 3: Trace a ReAct LoopTrace a full Reason → Act → Observe → Update State loop.Not done yet
- Activity 4: Search the Next StepCompare candidate actions, build a simplified search tree and choose the next step.Not done yet
- Activity 5: Verify or Reflect?Decide whether to verify a result, reflect on the strategy, ask for human review or simply stop.Not done yet
- Activity 6: Re-plan or Continue?Given a change in the environment or state, decide to continue the plan, re-plan, escalate to a human or stop.Not done yet
- AI 202 assessmentTwo levels: knowledge checks and scenario decisions.Not done yet
- Agent Planning & Reasoning CanvasBring everything together in an Agent Planning & Reasoning Canvas you can take back to work, research or an industry project.Not done yet
About this course
Outcomes
- Distinguish fixed workflows, agentic workflows and adaptive agents
- Decompose a goal into subgoals, actions, constraints and success criteria
- Explain and trace the ReAct Reason → Act → Observe loop
- Compare candidate actions and understand search trees and MCTS as agentic planning concepts
- Distinguish verification, evaluation and self-reflection
- Decide when to continue, re-plan, escalate or stop
- Complete an Agent Planning & Reasoning Canvas
Prerequisites
- AI 104 is recommended, or a basic grasp of agentic workflows, task decomposition, decision points and human-in-the-loop
- AI 201 is recommended, or an understanding of how agents use tools, APIs and MCP to reach external data and act
- The context, memory, knowledge and skills concepts from AI 103 will make the course easier
- No traditional AI algorithms or advanced programming background required
- The course focuses on designing agent planning and reasoning, not on mathematical derivation
Who it is for
- Learners who finished AI 104 and want agents to move from fixed processes to autonomous planning
- Staff and faculty who finished AI 201 and want to understand when and why an agent chooses different tools
- Coordinators and managers who plan agentic AI, automation, AI copilots or intelligent processes
- Students and researchers who want to go deeper into ReAct, search, self-reflection and adaptive planning
- Learners preparing to study multi-agent systems or agent harness engineering next
We suggest finishing AI 104 and AI 201 first. This is a suggestion only and never blocks you from starting.
Certificate
Earns a course completion certificate (L1); a capability certificate (L2) is also available through additional artifact review.
Requirements to earn it
- Complete all 7 course modules
- Pass the post-test assessment at 80% or higher
- Submit a project artifact and pass review
Course feedback
It takes 2 to 3 minutes and is used to improve the course.