Who the training is for
The programme is designed for beginners and mixed business, marketing, operations or technical profiles who want to understand agentic AI by building a real tool rather than accumulating theory.
It begins from a concrete problem, uses no-code first and keeps a person in the loop wherever an action can have a real consequence. It does not promise a magical autonomous agent or the replacement of an entire team.
How the programme works
You choose one evolving project at the beginning and improve it level by level. Each stage adds a practical capability and ends with a reusable deliverable.
The five possible tracks are a document copilot, email triage, lead pipeline, document extractor or content workshop. A personal project is possible when data is available, the scope is realistic and a person owns validation.
The five levels
Foundations distinguish analysis, creation and action. Prompting establishes robust instructions and expected outputs. The Agent level connects the model to data and tools. Safety adds permissions, traces, tests and human approval. Strategy measures time, cost, risk and the business case.
Every level is applied to the same project, so the final outcome is one progressively stronger tool rather than five disposable exercises.
Deliverables and prerequisites
The intended deliverables include a use-case brief, structured prompts, a workflow, an action-permission matrix, a representative test set, a cost and value estimate and a presentable demonstration.
Programming experience is not required. You need a computer, internet access and the willingness to test, correct and document. A dependable agent is not created by a single prompt.
Status, without an invented launch date
The training is in preparation. The priority list is used to share confirmed opening information and final conditions. No date, price, certification or availability is claimed before it is real.
Sensitive actions such as sending a message, changing business data, committing expenditure or publishing content remain subject to explicit human validation until the system has demonstrated that it meets its criteria.