Issued by Foundryy · Generative AI
Agentic AI and AI Agents
Learn to build AI agents that plan, call tools and complete multi-step tasks, and to keep them reliable and safe. You will implement the tool-calling loop yourself, add memory and guardrails, and test an agent that automates a real workflow.
- Duration
- 3 months
- Study time
- 96 hours
- Modules
- 7
- Cost
- At enrollment
- AI Agents
- Tool Calling
- LLM APIs
- Python
- Agent Evaluation
Curriculum
- 18 h
What makes a system agentic
Workflows versus agents, When not to use an agent
- 216 h
The tool-calling loop
Defining tools, Planning and acting, Handling tool errors
- 312 h
Memory and context
Short- and long-term memory, Context limits
- 412 h
Guardrails and safety
Permissions and human approval, Prompt injection, Spending limits
- 512 h
Multi-step and multi-agent patterns
Task decomposition, Handoffs between agents
- 612 h
Evaluating agents
Scenario tests, Reading traces, Regression checks
- 724 h
Capstone
Choose a workflow, Build and test, Demo and failure review
Who can enroll
Comfortable Python and some experience calling an API. Taking Generative AI Foundations first is recommended.
To earn the certificate
- Complete every module of the curriculum
- Submit a capstone project: a tool-using agent that completes a multi-step workflow, with logged traces, guardrails and a test suite of at least 15 scenarios
- Pass a final review by a Foundryy reviewer
- Follow Foundryy’s academic-integrity policy: the work you submit must be your own
Issued by Foundryy after the team has reviewed your capstone against the requirements.