Catalog AI Agents & Agentic Systems
From beginner to expert — buy individually or as a bundle with a certificate of completion.
AI agent roles involve designing systems that can understand an objective, use tools, access data, make controlled decisions and execute actions autonomously or under supervision. This learning path develops skills used by AI Engineers, Agentic Systems Engineers, Automation Engineers and AI Architects to design, orchestrate, secure, evaluate and deploy reliable, observable and controlled agents in production. By the end of the learning path, you will be able to scope an agentic use case, design instructions and tools, manage memory and state, connect MCP and external services, orchestrate multiple agents, secure sensitive actions, keep humans in the loop, implement testing, tracing and monitoring, and deploy a professional end-to-end agentic system.
Each course brings together 8 interactive lessons in a Learning Workspace designed to turn knowledge into operational skills. Guided lessons, dynamic quizzes, conversational support, practical cases, additional resources and progress tracking let you learn effectively, at your own pace. The 15 courses follow a level-by-level progression, from scoping a first AI agent to designing, securing and deploying a complete, observable multi-agent system.
🎓 15 courses • 📚 120 interactive lessons • ⏱️ 45 to 75 hours of conversational learning
⭐ Best value
AI Agents & Agentic Systems — Complete Pack (15 courses)
AGENTS01 → AGENTS15 · 1 800 credits included
142,90 € excl. VAT
Savings : 104,60 €
29,70 €
22,90 excl. VAT
Savings : 6,80 €
Understand AI agents and scope a use case
distinguish assistant, workflow and agent, then decide whether agentic autonomy actually adds value
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
No specific prerequisite to start. Confidence with digital tools is recommended; later technical courses may use development or data concepts.
- Understand what sets apart assistant, workflow and agent
- Understand the perceive, decide, act, observe loop
- Assess the value, cost, latency and risk of autonomy
- Workshop: decide whether a need truly justifies an agent
- Distinguish assistant, workflow and agent, then decide whether agentic autonomy actually adds value
- Identify goal, environment, tools and state
- Compare LLM-driven decisions and deterministic logic
- Workshop: decide whether a need truly justifies an agent
3 to 5 h
9,90 € excl. VAT
Design an agent's instructions and behavior
define agentic behavior that is predictable, testable and bounded by explicit rules
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Define the agent's role, goal and responsibility
- Separate instructions, context and user data
- Handle uncertainty, refusal and clarification requests
- Workshop: turn a vague prompt into a testable behavioral contract
- Define agentic behavior that is predictable, testable and bounded by explicit rules
- Structure constraints and standing rules
- Design action policies and stopping criteria
- Workshop: turn a vague prompt into a testable behavioral contract
3 to 5 h
9,90 € excl. VAT
Give an agent tools and structured actions
let an agent act on external systems with typed, validated and controlled tools
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Understand the tool-calling cycle
- Define reliable parameters, descriptions and schemas
- Validate tool arguments, outputs and errors
- Workshop: build an agent using several controlled tools
- Let an agent act on external systems with typed, validated and controlled tools
- Turn a function into a typed tool
- Connect a SQL database while limiting allowed operations
- Workshop: build an agent using several controlled tools
3 to 5 h
9,90 € excl. VAT
38,70 €
28,90 excl. VAT
Savings : 9,80 €
Manage context, state, memory and sessions
maintain an agent's continuity without confusing conversation history, business state and persistent memory
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Distinguish run context, history and business state
- Design short-term memory
- Separate user data, preferences and secrets
- Workshop: design a memory and persistence strategy
- Maintain an agent's continuity without confusing conversation history, business state and persistent memory
- Understand sessions and multi-turn continuity
- Manage compaction and context limits
- Workshop: design a memory and persistence strategy
3 to 5 h
12,90 € excl. VAT
Connect an agent to reliable knowledge
give agents external information that is traceable, fresh and suited to the need
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Distinguish model knowledge from external knowledge
- Use search and dynamic retrieval
- Control freshness, rights and source quality
- Workshop: build an agentic answer grounded in traceable sources
- Give agents external information that is traceable, fresh and suited to the need
- Connect relevant files and documents
- Preserve provenance and citations
- Workshop: build an agentic answer grounded in traceable sources
3 to 5 h
12,90 € excl. VAT
Connect tools and data with Model Context Protocol
use MCP to standardize how agentic applications access external tools, resources and systems
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Understand MCP's role in an agentic system
- Understand transports, sessions and the stateless core
- Connect an MCP server to an agent
- Workshop: build and secure a small MCP integration
- Use MCP to standardize how agentic applications access external tools, resources and systems
- Distinguish MCP client, MCP server, tools and resources
- Design permissions and authorization for MCP tools
- Workshop: build and secure a small MCP integration
3 to 5 h
12,90 € excl. VAT
47,70 €
35,90 excl. VAT
Savings : 11,80 €
Design durable agentic workflows
combine LLM decisions and deterministic control in workflows that can wait, resume and handle errors
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Distinguish agentic loop from explicit workflow
- Represent and evolve workflow state
- Make actions idempotent and replayable
- Workshop: build an agentic workflow that resumes after a failure
- Combine LLM decisions and deterministic control in workflows that can wait, resume and handle errors
- Design sequences, branches and loops
- Handle timeouts, retries and partial errors
- Workshop: build an agentic workflow that resumes after a failure
3 to 5 h
15,90 € excl. VAT
Orchestrate several specialized agents
choose and implement multi-agent patterns without needlessly multiplying models and costs
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Understand why to specialize several agents
- Delegate with a handoff
- Parallelize then aggregate subtasks
- Workshop: compare a single-agent and a multi-agent solution
- Choose and implement multi-agent patterns without needlessly multiplying models and costs
- Use a specialized agent as a tool
- Route a task to the right agent
- Workshop: compare a single-agent and a multi-agent solution
3 to 5 h
15,90 € excl. VAT
Make agents interoperate with A2A
understand and implement communication between independent agents using an open protocol
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Understand the role of the A2A protocol
- Understand tasks, messages and artifacts
- Delegate a task and retrieve its result
- Workshop: build a delegation between two independent agents
- Understand and implement communication between independent agents using an open protocol
- Discover AgentCard and published capabilities
- Discover and select a remote agent
- Workshop: build a delegation between two independent agents
3 to 5 h
15,90 € excl. VAT
56,70 €
42,90 excl. VAT
Savings : 13,80 €
Secure an agentic system
reduce the risks created by agents able to read data and execute actions
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Build a threat model for an agent
- Limit tool permissions with least privilege
- Reduce exfiltration, SSRF and unauthorized access
- Workshop: audit and harden an agentic system
- Reduce the risks created by agents able to read data and execute actions
- Understand direct and indirect prompt injection
- Validate arguments and results before acting
- Workshop: audit and harden an agentic system
3 to 5 h
18,90 € excl. VAT
Put a human in the loop and control sensitive actions
have risky actions approved, modified or declined without breaking workflow continuity
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Classify actions by risk level
- Pause a run before a sensitive operation
- Serialize state and resume after a decision
- Workshop: build a complete human-in-the-loop workflow
- Have risky actions approved, modified or declined without breaking workflow continuity
- Define which actions require approval
- Approve, modify or decline an action
- Workshop: build a complete human-in-the-loop workflow
3 to 5 h
18,90 € excl. VAT
Test, evaluate, trace and diagnose agents
measure an agent's quality on reproducible scenarios and understand why a run succeeds or fails
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Define the behaviors to evaluate
- Test final answers and structured outputs
- Automate regression tests
- Workshop: diagnose an agentic regression from traces
- Measure an agent's quality on reproducible scenarios and understand why a run succeeds or fails
- Build a dataset of representative scenarios
- Measure success, cost, latency and step count
- Workshop: diagnose an agentic regression from traces
3 to 5 h
18,90 € excl. VAT
74,70 €
48,90 excl. VAT
Savings : 25,80 €
Make agents reliable, performant and cost-controlled
design agents able to run over time despite errors, quotas, latency and variable costs
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Define turn, token, cost and latency budgets
- Control idempotence and partial actions
- Design fallback and degraded modes
- Workshop: stabilize a slow, costly and fragile agent
- Design agents able to run over time despite errors, quotas, latency and variable costs
- Handle timeouts, retries and backoff
- Set up caching and result reuse
- Workshop: stabilize a slow, costly and fragile agent
3 to 5 h
24,90 € excl. VAT
Build and deploy a professional agentic system
assemble tools, memory, MCP, specialized agents, approvals, guardrails, tracing and an API into an operable system
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Structure an agentic Python application
- Add memory, sessions and state storage
- Add guardrails and human approvals
- Workshop: deploy a reproducible, observable agentic system
- Assemble tools, memory, MCP, specialized agents, approvals, guardrails, tracing and an API into an operable system
- Build agents, tools and structured outputs
- Add specialized agents and controlled handoffs
- Workshop: deploy a reproducible, observable agentic system
3 to 5 h
24,90 € excl. VAT
Full professional AI Agents project
design, secure, evaluate, deploy and defend an agentic system end to end
Course information
AI Engineers, developers, architects and automation specialists who want to design, secure and operate agentic systems.
A working knowledge of the foundations covered earlier in the pathway is recommended. This course builds on concepts introduced in previous courses.
- Scope the goal, perimeter and limits of autonomy
- Connect data, APIs and MCP
- Create an evaluation dataset, tests and tracing
- Final workshop: demonstrate and defend the complete agentic system
- Design, secure, evaluate, deploy and defend an agentic system end to end
- Design the architecture, instructions and tool schemas
- Build the threat model, permissions and approval matrix
- Final workshop: demonstrate and defend the complete agentic system
3 to 5 h
24,90 € excl. VAT