Advanced · Self-paced learning
Inquiry This course is not yet bookable.
In a prepared practice environment, you specify which data and tools the agent may use for this task. Run a case: the agent reads an approved discrepancy record and prepares a draft. You review and approve it as the caseworker before the environment carries out the next step as a simulated action.
Then test what happens when the agent attempts an unauthorised change to supplier data or approval arrives too late. Inspect the log, adjust permissions or approval rules as needed and repeat the test. Finally, apply what you have learned to a process in your organisation: what should the agent do, what information does it need and which decisions remain with people?
A local tool for testing access permissions with synthetic data is planned as a course companion. Run permitted and prohibited actions, compare expected and actual results and keep your configuration. The course also explains which access boundaries this tool enforces and where additional isolation is needed.
Take away a runnable agent setup with your configuration, example data, start instructions and executed test cases for further use. The operating sheet records permitted actions, access limits and intervention routes.
You are responsible for a process or an AI initiative and can outline its objective, data access, roles and limits. Access to relevant internal process and approval information is helpful; where access is not approved, you work on a case study with a comparable decision situation. No programming knowledge is required.
Planned self-study with 8–10 hours of work. You use practice data and simulated tools in a prepared environment, with guidance and explanatory self-checks.
Planned learning time: 8–10 hours
