This 3-day (18-hour) training course is designed for beginners who want to understand how intelligent agents work and learn to build, step by step, a first AI agent capable of interacting with a large language model (LLM), using tools and carrying out tasks semi-autonomously.
Training Objectives
By the end of this training, you will be able to:
- Understand generative AI and large language models (LLMs): ChatGPT, Gemini, Claude.
- Tell the difference between a classic chatbot and a true autonomous AI agent.
- Master the core concepts: prompt engineering, memory, tools, reasoning.
- Connect to an AI model through an API (OpenAI, Anthropic, etc.).
- Design an agent workflow tailored to a concrete business need.
- Build and test a working AI agent from start to finish.
- Identify the limitations, risks and security best practices of AI agents.
Day 1 – Understanding AI Agents
The first day lays the theoretical and practical foundations needed to understand modern AI agents.
- Introduction to generative artificial intelligence and LLMs.
- How an AI agent works (architecture: model, instructions, memory, tools, actions).
- Overview of the main professional use cases (customer support, data analysis, automation).
- Introduction to prompt engineering: structuring your instructions to get the best answers.
- Getting to know the technical environment.
- First hands-on exercise with an AI model.
Day 2 – Building an AI Agent
The second day is entirely devoted to hands-on development. Participants build their first working AI agent.
- Connecting to a language model through an API (OpenAI GPT-4 / Anthropic Claude).
- Writing the system instructions and defining the agent's role.
- Adding tools and functions (function calling, external API calls).
- Managing context and conversational memory.
- Setting up an agentic workflow (chain of thought, ReAct).
- Testing, debugging and improving responses.
Day 3 – Hands-On Project, RAG and Best Practices
The third day wraps up the training with a complete mini-project and an introduction to advanced techniques.
- Designing an agent that addresses a concrete need of the participant.
- Connecting to documents or a data source (PDFs, databases).
- Introduction to RAG (Retrieval-Augmented Generation): connecting the agent to your own data.
- Setting up control rules, validation and guardrails.
- Security, privacy and API cost control.
- Evaluating the agent's performance.
- Presentation and validation of the final project.
🎓 Training available on-site (Sousse, Tunis) or remotely. Flexible format: 2 days (accelerated), 3 days (standard), or 4 days (with full deployment and advanced RAG). Eligible for TFP (Professional Training Tax) funding.
Included Deliverables
Each participant leaves with:
- Complete course materials (slides + reference guide).
- Source code and commented hands-on examples.
- Additional resources (articles, documentation, recommended tools).
- The AI agent mini-project built during the training.
- Certificate of attendance (Tunisian State-accredited training).
Join the Training
To organize this training in your company or institution, contact Houssem Ankoud directly. Please specify the number of participants, the preferred format (on-site/remote), the team's technical level, and the dates you have in mind. Reply guaranteed within 24 hours.

