An AI agent is something different from an assistant you ask questions: it is a system given an objective, which decides on its own which steps to take and which tools to use to get there.
The course approaches agent design practically: when an agent is warranted and when a linear automation is enough, how to define the tools it can access, how to handle errors and the cases where the agent gets it wrong.
We work with concrete tools such as n8n, Flowise and the main model APIs, building agents for real cases: automating repetitive work, handling recurring requests, integrating with company systems. Security and governance have a dedicated module, because an agent acting on company systems needs supervision.
The observe-decide-act loop, the concept of a tool, and the practical limits of these systems' autonomy.
An overview and comparison of the platforms, with a guided build of a first working agent.
The use cases that pay back soonest: routing, classification, extracting data from documents, compiling reports.
Answering recurring requests from company documentation, with escalation to a person when needed.
Connecting the agent to the systems holding the data, with attention to permissions and what it can actually change.
Action limits, logging, human review and intervention plans. An agent operating on live systems has to stay supervised.
No course is a fixed package. Duration, angle and depth are agreed around the objectives of the people taking part, and topics from different courses can be combined into a single programme. I work with corporate groups and with one person alone: tell me what you need and we will build the programme.
Request informationThe tools used are low-code, but the course is technical. People with development or systems integration experience get more from it.
Automations follow a fixed path we define. An agent decides its own steps. The course also helps recognise when the first option is preferable.
In corporate programmes yes, subject to access permissions. Otherwise we use a test environment.
n8n and Flowise have free or self-hosted versions. Model APIs are pay-as-you-go, at modest cost for the exercises.