Generative AI & assistants
Build your first AI agent
Design, build and secure an agent that can reason, use tools and act.
- Duration
- 14 h · 2 days
- Level
- Intermediate
- Format
- On site · Remote · Blended
- Max. group size
- 12 participants
Overview
This program guides participants from understanding the AI agent concept to putting a first useful agent into service. It distinguishes workflows from agents, designs the scope and guardrails, then builds an agent with tools, memory and a knowledge base. A large part is devoted to testing, security and human oversight, which are essential for an agent that acts.
Who this program is for
- Professionals who have already used generative AI and want to take action
- Project managers, consultants and no-code/low-code developers
- Teams wanting to prototype an internal assistant
Prerequisites
- Having completed a prompt engineering program or mastering its basics
- Comfort with an automation tool (Make, n8n or equivalent) is desirable
- No programming required
Learning objectives
- Tell a chatbot, an automated workflow and an agent apart
- Define a use case, a scope and guardrails
- Build an agent equipped with tools and memory
- Connect the agent to documents and external services
- Test, evaluate and secure its behavior
- Deploy the agent and monitor its activity
Detailed curriculum
6 modules · 14 h · 2 days
Day 1
Understanding AI agents
70% theory · 30% practice
- Model, tools, memory, reasoning loop
- Chatbot, workflow or agent: which to choose
- Examples of agents in business
- Current limits and risks
Designing your agent
40% theory · 60% practice
- Choosing a high-impact use case
- Defining scope, goals and success criteria
- Listing the tools needed
- Planning guardrails and human validation
🛠 Hands-on workshops
Agent design sheet1 h
Each participant fills in the design sheet of their future agent.
✅ Deliverable : A complete design sheet
Building an agent without code
20% theory · 80% practice
- Creating the agent in a no-code tool
- Writing its instructions and role
- Connecting a first tool
- First tests and adjustments
🛠 Hands-on workshops
First operational agent2 h
Build an agent that meets a simple need end to end.
✅ Deliverable : A testable agent
Day 2
Tools, memory and knowledge
20% theory · 80% practice
- Tools: search, spreadsheets, email, calendar
- Short-term and long-term memory
- Giving access to reference documents
- Managing access and permissions
🛠 Hands-on workshops
Equip the agent with several tools1.5 h
Add tools and a knowledge base to the day 1 agent.
✅ Deliverable : A multi-tool agent
Test, evaluate and secure
30% theory · 70% practice
- Building a set of test scenarios
- Spotting hallucinations, loops and poor decisions
- Limiting risky actions
- Human validation and logging
Deploy, monitor and final project
15% theory · 85% practice
- Going live and access channels
- Monitoring costs and quality
- Improving the agent over time
- Agent presentations
🛠 Hands-on workshops
Final presentation1.5 h
Each participant presents their agent, its guardrails and its test results.
✅ Deliverable : A demonstration and a test set
Tools and technologies
ChatGPTClaudeGoogle SheetsMaken8n
Teaching methods
- Demonstrations and guided step-by-step building
- Hands-on workshops on a running case
- Group security and behavior reviews
- Pair work
Assessment
- Exercises in each module
- Presentation of the agent built and its test set
- Knowledge-check quiz
Required equipment
- A computer with an internet connection
- Accounts on the automation tool and AI model used (free or trial plans)
Resources provided
- Course materials and agent templates
- Agent design sheet
- Test and security grid
Career outcomes
- A first working agent for your own activity
- A reusable method for designing other agents
- Foundation for the RAG and LangChain programs