Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 14 hours
Course Outline
- Day 1: Agent Planning and Initial Construction
- From Assistants to Agents
- Comparing chatbots, assistants, agents, and automations
- Capabilities that agents can reliably perform today
- Case studies from administration, procurement, HR, and customer service
- Activity: Categorize three personal workflows by determining whether a chatbot, assistant, or agent is the appropriate fit.
- Identifying the Right Task
- Evaluation criteria including frequency, defined steps, and verifiable outcomes
- Describing inputs, processing logic, outputs, and control points
- Establishing clear success metrics
- Activity: Complete an agent profile for your chosen workflow.
- Core Components of an Agent
- Defining goals and instructions
- Integrating knowledge from documents and file storage
- Connecting tools such as email, calendars, spreadsheets, and web sources
- Setting limits and escalation paths
- Activity: Draft instructions and boundaries for your agent.
- Creating the First Agent
- Selecting tools based on corporate approvals
- Implementation using Copilot Agent Builder, custom GPTs, Claude projects with skills, or Langdock agents
- Testing with three practical scenarios
- Activity: Build your first agent and validate it with three test cases.
- Outcome: Each participant will possess an agent profile and a first tested agent for a personal workflow.
- From Assistants to Agents
- Day 2: Agents with Tools, Workflows, and Control Mechanisms
- Agents That Execute Actions
- Moving from response generation to task execution
- Connecting to email, file storage, CRM systems, and spreadsheets
- Browsers as an extension for agent capabilities
- Activity: Connect the Day 1 agent to a tool, such as file storage or a spreadsheet.
- Integrating Agents into Workflows
- Configuring triggers such as new emails, form submissions, or scheduled times
- Incorporating AI agent steps in n8n or Make
- Storing results in spreadsheets, ticket systems, or mailboxes
- Activity: Construct a workflow where an agent classifies incoming requests and saves a draft.
- Human Oversight
- Establishing approval checkpoints before sending, booking, or deletion actions
- Maintaining logs of agent activities
- Protocols for handling uncertainty
- Activity: Integrate an approval step via email or Teams into the workflow.
- Testing and Operation
- Defining test cases with expected outcomes and failure scenarios
- Managing costs, usage limits, and maintenance
- Assigning responsibilities and creating documentation
- Activity: Execute five test cases and produce concise operational documentation.
- Presentation and Next Steps
- Sharing results with the group
- Selecting the next agent candidate
- Clarifying responsibilities for maintenance and future development
- Activity: Present your agent in a three-minute summary.
- Outcome: Each participant will have an agent equipped with a tool, an approval checkpoint, test cases, and concise operational documentation.
- Agents That Execute Actions
Requirements
- Access to at least one corporate agent tool (such as Copilot Agent Builder, ChatGPT, Claude, Langdock, or n8n) and a recurring workflow from your own domain.
- No programming experience is required.
Target Audience
- Specialist users, key users, and process owners looking to delegate recurring tasks to AI agents.
Testimonials (1)
real life examples
