Designed for managers, academic and professional services staff, business owners and professionals who use, or want to use, generative AI in their work.

The programme is useful for anyone who wants more consistent results from AI when drafting, summarising, extracting actions, producing briefings, planning or supporting other everyday workplace tasks.

No programming or previous prompt engineering experience is required. Basic digital literacy is sufficient.

This programme will help you if you want to:

• Write clearer prompts that produce more relevant outputs
• Give AI the right context, source material, constraints and output format
• Break more complex tasks into manageable stages
• Use examples and iterative refinement effectively
• Create reusable prompt templates instead of starting from scratch each time
• Check AI outputs for fabricated facts, omissions, calculations and bias
• Compare prompt versions using consistent quality criteria
• Build human review and data safeguards into your AI workflows

Effective prompting is less about finding a magic phrase and more about designing a clear, testable workflow. This course focuses on practical techniques you can apply across tools and workplace tasks, while keeping human responsibility for the final output.

Throughout the day, you will move from a baseline prompt to increasingly structured approaches, test outputs against supplied source material and finish by developing a reusable prompt for a task relevant to your own role.

Understand how prompts shape outputs

Start with the foundations: how prompts, context and model variability influence results. You will complete a baseline task and consider responsible tool use, including the handling of sensitive information.

Write clearer, more effective prompts

Learn to specify the objective, audience, context, source material, constraints and output format. You will explore when roles are useful, when the model should ask for clarification and how small changes can improve an otherwise vague request.

Build reusable workflows

Compare zero shot and few shot prompting, use examples effectively and separate instructions from source text. You will also break complex tasks into stages with review points rather than expecting one prompt to do everything.

Ground and check outputs

Work from supplied documents, ask for source references and flag missing evidence. You will practise checking fabricated facts, calculations, omissions and bias, and learn why instructions contained inside source documents should be treated cautiously.

Test and refine prompts systematically

Compare prompt versions against the same source pack and evaluation rubric. This makes improvement measurable and helps you understand whether a change has actually improved accuracy, relevance, completeness or instruction compliance.

Apply prompt engineering to a real workplace task

Choose a task such as an options appraisal, project plan, policy briefing or learning resource. Define success criteria, build and refine the prompt, test it with two input examples and record the human checks that remain necessary.

Learn through peer review

Review outputs for accuracy, relevance, completeness, instruction compliance and responsible use. Compare the final result with your baseline task, explain what changed and identify what still requires human verification.

The emphasis is on useful outputs that you can take back to work and continue improving, rather than a collection of abstract prompting techniques.

You will leave with:

• A small prompt portfolio covering drafting, summarising and action extraction
• One reusable workplace prompt template tested on two input examples
• A prompt evaluation and revision record
• A checklist for evidence checks and responsible use
• A 30 day application plan with a quality and efficiency measure

By the end of the programme, you should be able to:

• Explain how prompts and context influence AI outputs and recognise model limitations
• Write clear prompts specifying task, audience, context, constraints and output format
• Apply examples, task decomposition and iterative refinement to workplace tasks
• Use supplied source material and verify factual accuracy, completeness and relevance
• Test prompt versions against a consistent rubric and record improvements
• Create reusable prompt templates with appropriate data safeguards and human review

Want the full programme detail?

Download the full course outline for the complete timetable, learning outcomes, activities, assessment approach and supporting materials.

Course duration

One day, 09:00 to 17:00.

The programme provides 6.5 guided learning hours, excluding lunch and refreshment breaks.

Participants

Suggested cohort size: 15 to 25 participants.

The group format supports guided practice, paired review, individual application and tutor feedback.

How the day is delivered

The programme is practical and tool neutral. Exercises use fictional or public inputs, with human responsibility retained for final outputs.

• Short demonstrations
• Guided practice
• Hands on workshops
• Paired review
• Individual workplace challenge
• Tutor feedback and knowledge check

Morning: From foundations to reusable workflows

The morning builds the core techniques for giving AI clearer instructions and structuring more reliable workflows.

• Prompt foundations and baseline task
• Writing effective prompts
• Using context, source material, constraints and output formats
• Zero shot and few shot prompting
• Examples, task decomposition and review points

Afternoon: Check, refine and apply

The afternoon focuses on grounding outputs in evidence, testing prompt versions and applying the techniques to a workplace task.

• Grounding and checking outputs
• Detecting fabricated facts, omissions and bias
• Prompt comparison using a consistent rubric
• Applied prompt engineering challenge
• Peer review and revision
• 30 day personal action plan

Preparation and equipment

Participants should bring a non sensitive workplace task. A laptop, internet access and access to an organisation approved generative AI tool are required.