Course Objective & Scope
Agents most often fail at planning long tasks and updating that plan as conditions change. The programme examines planning approaches such as ReAct, plan-and-execute, hierarchical task decomposition, reflection and self-critique, and the role of reasoning models like the OpenAI o-series, Claude extended thinking and Gemini thinking in these loops.
Participants build agents that solve the same task with different planning strategies, compare success rate, step count and cost, and produce a loop design template covering replanning and termination rules.
Who Should Attend?
- AI engineers developing agents
- Solution architects designing complex automation scenarios
- Technical teams responsible for improving agent performance
- Prerequisite: Python and basic agent development experience
Key Takeaways
- Select the right planning strategy for the task type
- Build planners that decompose complex tasks into subtasks
- Recover from errors through reflection and replanning
- Use reasoning models while balancing cost and latency
- Write termination rules that prevent infinite loops and wasted steps
Curriculum
Fundamentals of the Agent Loop
- The observe–think–act loop and ReAct
- Single-step versus multi-step tasks
Planning Strategies
- Plan-and-execute and hierarchical task decomposition
- Plan representation: list, graph and task tree
Reasoning Models and Thinking Budgets
- o-series, Claude extended thinking and Gemini thinking
- Choosing different models for planner and executor
Reflection and Replanning
- Self-critique, verifier steps and error recovery
- When should the agent replan?
Termination and Resource Control
- Success criteria, step and token limits
- Loop detection and handing over to the user
Comparative Lab
- Measuring success, cost and speed of strategies on the same task
- Loop design template
Other Courses in This Category
-
Agent memory and context management for long-horizon tasks and multi-session interactions: working memory, episodic and semantic long-term memory, context compaction, and persistent memory architecture with Mem0 and LangGraph Store.
- Live Course
- 5 modules
-
Multi-agent architectures that split complex work across specialised agents: supervisor, hierarchical and swarm patterns, comparing LangGraph, CrewAI and AutoGen, the A2A protocol and shared state management.
- Live Course
- 5 modules
-
Human-in-the-loop agent architecture that preserves human oversight on critical actions: interrupt/resume flows, risk-based approval thresholds, escalation queues and agents that improve from human feedback.
- Live Course
- 5 modules
-
Connecting enterprise databases, APIs and internal systems to LLMs and agents cleanly and securely with Model Context Protocol (MCP): MCP server/client development, authorisation and enterprise integration architecture.
- Live Course
- 5 modules
-
Turning institutional knowledge, procedures and tools into modular, reusable agent skills: SKILL.md structure, progressive disclosure, templates, scripts and enterprise skill library governance.
- Live Course
- 5 modules
-
End-to-end and step-level evaluation of multi-step agent systems: trajectory evals, tool calling accuracy, task completion metrics, simulated user/environment testing and CI/CD eval pipelines.
- Live Course
- 5 modules
Get in touch
