Course Objective & Scope
Leaving agents completely unattended on irreversible actions such as payment approval, customer communication, record deletion or contract dispatch creates severe operational risk. This programme covers human-in-the-loop (HITL) architectures that embed human oversight into the agent loop without creating bottlenecks.
Participants classify low- and high-risk actions, build a prototype using LangGraph interrupt/resume and asynchronous webhooks that pauses execution and resumes after human approval or edits, and prepare an approval matrix and checklist for internal use.
Who Should Attend?
- Process and product owners responsible for agent initiatives
- Software and AI engineers building enterprise workflows and agents
- Risk, compliance and operations teams
- Prerequisite: basic agent concepts; intermediate Python/API familiarity for labs
Key Takeaways
- Classify tool calls by reversibility and business impact
- Build asynchronous approval flows that persist state until a human responds
- Handle approve, edit-arguments and reject/redirect scenarios cleanly
- Design operator notifications that present concise evidence and context
- Turn human corrections into evaluation datasets and few-shot examples
Curriculum
Autonomy Spectrum and Risk Classification
- Read-only, drafting and state-mutating external actions
- Dynamic approval thresholds by amount, confidence score and segment
Interrupt and Resume Architecture
- Freezing state with checkpointers and managing timeouts
- Collecting asynchronous approvals via Slack, Teams or internal portals
Intervention Patterns: Approve, Edit, Redirect
- Executing tool calls with human-edited arguments
- Feeding rejection feedback back to the agent for replanning
Preventing Approval Fatigue and Operator UX
- Evidence summaries, diff views and one-click decision cards
- SLA tracking, fallback escalation and audit logging
Learning Loops from Human Feedback
- Capturing and analysing human edits and overrides
- Improving prompts, rules and eval datasets from approval history
Other Courses in This Category
-
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
-
Securing tool-using agents that act on external systems: indirect prompt injection defence, delegated user authorisation with OAuth/OBO, sandbox isolation, guardrails and OWASP Agentic security controls.
- Live Course
- 5 modules
-
Foundational programme on how LLM-based agents work: model, tools, memory and the control loop; how agents differ from chatbots and workflow automation, and how to pick the right first use case for your organisation.
- Live Course
- 6 modules
-
Agentic AI for executives: impact on business models, processes and organisation, function-by-function use cases, investment and ROI assessment, operational risk, governance and redesigning teams that work alongside agents.
- Live Course
- 5 modules
Get in touch
