YOUR PROJECT OUTCOME
A scoped AI prototype, at least ten test cases, an error register and a dossier covering human oversight, responsibilities and release criteria.
Who it is for
Marketing leaders, consultants and small teams experienced in using AI assistants and interested in repeatable processes.
What you need to start
Familiarity with generative AI and a willingness to begin with demonstration data. Production integrations require a dedicated technical assessment.
The curriculum
01Choosing the process
+
Map tasks, time and errors, then select a bounded use case. Use AI to compare three approaches, including a simpler manual procedure, and verify costs and dependencies.
02Sources, memory and instructions
+
Organise an authorised knowledge base and separate data from instructions. Test source retrieval, updates and resistance to external content attempting to change the rules.
03Permissions and approvals
+
Separate the agent’s reading, proposing and acting capabilities. Define permitted tools, minimum data and stages requiring explicit human approval.
04The prototype and its workflow
+
Build a prototype using demonstration data with defined inputs, outputs and logs. Document integrations and controls, retaining a manual route for unreliable steps.
05Evaluation and error handling
+
Run normal, ambiguous and adversarial cases. Measure correctness, completeness, time, cost and human intervention; improve the process and repeat relevant tests.
06From prototype to release
+
Present the prototype, findings and limitations. Define responsibilities, suspension and a manual fallback, identifying technical or regulatory checks for qualified professionals.
The final project
An AI assistant prototype for content or briefing work, with at least ten tests, a manual-process comparison, an error register and a human-oversight plan.
Planned assessment criteria
Planned review criteria include process usefulness, control quality, data handling and behaviour in critical cases. The programme does not confer regulatory compliance or security certification.
Tools and materials
- AI assistant
- n8n or an equivalent prototyping tool
- Demonstration data
- Test and approval register
Listed tools are examples, not partnerships. Access requirements and possible additional costs will be specified before enrolment.
← Back to the course catalogue