AI Accelerator:
From Zero to Confident

Complete course resources

Use the slide presentation for delivery, the facilitator manual for detailed scripts, the participant workbook for exercises and the assessment rubric for completion checks. All resources are included in the downloadable ZIP package.

Practical result: every learner creates a reusable assistant, runs five tests, and documents one improvement.

1. Course outcomes

By the end of the session, participants should be able to:

• Explain generative AI and describe common limitations.
• Use a repeatable prompt structure.
• Apply AI to communication, research and organisation.
• Identify hallucination, bias and privacy risks.
• Create and test a reusable AI assistant for a specific task.

Preparation checklist

• Prepare a projector or screen share.
• Ask participants to bring a device and access to an AI chat tool.
• Have a no-login demo or paired-device option for participants who cannot sign in.
• Tell participants not to bring real confidential records or personal data into practice.
• Prepare one demonstration task, such as turning rough notes into a professional email.

2. Run of show — 210 minutes

3. Teaching notes and activities

Module 1 — Understanding AI (25 min)

Explain: AI systems identify patterns and perform tasks such as generating text, summarising information, classifying content and assisting with analysis. Generative AI creates new outputs, but can produce incorrect or unsupported statements.

Ask: “What is one repetitive task you would like to make easier?” Write down examples and reuse them in later exercises.

Activity: Participants ask their AI tool, “Give me five ways AI could help someone in my role.” They then add their role, audience and a recurring task. Compare the results.

Module 2 — The art of the prompt (35 min)

Teach this five-part pattern: role, task, context, format and quality check. Explain that not every prompt needs every part, but adding relevant detail usually improves results.

You are [role].
Help me [specific task].
Context: [background, audience and goal].
Constraints: [tone, length, deadline, rules].
Format: [email, table, checklist, plan].
Before finishing, check [quality criteria] and flag uncertainty.

Activity: Improve “Help me promote my business.” Ask participants to specify the business or service, target audience, location, budget, objective, output format and tone. Have them run both versions and compare.

Module 3 — Everyday AI (35 min)

Demonstrate three short tasks: turn bullet points into an email; explain a concept in beginner language and generate quiz questions; convert rough notes into a table of actions, owners and deadlines.

Activity: Participants choose a low-risk task, ask for a first output, request an improvement, then request a different format. Ask what the human needed to correct.

Module 5 — Responsible AI (20 min)

• Hallucination: fluent output may contain false details, invented references or unsupported claims.
• Bias: outputs may reflect stereotypes or unfair assumptions.
• Privacy: avoid entering passwords, financial details, health records, client information or confidential company data unless the service and relevant policies explicitly permit it.
• Verification: independently check important facts, especially current, legal, medical, financial or safety-related claims.

Facilitator rule: Use fictional or anonymised information in exercises. AI can assist professionals, but it does not replace required professional judgement, approval or accountability.

4. Detailed lab: create an AI assistant (65 min)

Goal: Each participant creates a small assistant for one well-defined task. Examples include an email drafting coach, meeting action tracker, customer FAQ helper, lesson planner, social media content planner or personal study coach.

Step A — Define the job (5 min)

Have participants complete this sentence:

My assistant helps [specific user] do [specific task] by producing [useful output].

Encourage narrow use cases. “Help me with everything” is too broad. “Turn meeting notes into a concise action list with owners and due dates” is testable.

Step B — Write the instructions (10 min)

Participants customise the following prompt blueprint. Replace every bracketed item. Remove instructions that do not apply.

NAME:
[Give the assistant a clear name.]

IDENTITY:
You are [role or type of assistant] supporting [intended user].

MISSION:
Your job is to help me [one specific task and desired outcome].

WORKFLOW:
1. Identify what the user wants to achieve.
2. If essential information is missing, ask concise clarifying questions.
3. Work only with information provided or sources that you can actually access.
4. Produce a useful first draft or result.
5. Check the result against the requirements before responding.

OUTPUT:
Use [format, e.g. checklist, table, email or step-by-step plan].
Use a [tone] tone and [level of detail].
Include [required fields or sections].

RULES AND BOUNDARIES:
- Do not invent facts, names, dates, prices, policies or sources.
- Clearly label assumptions and uncertainty.
- Use placeholders when details are missing.
- Do not request or expose unnecessary personal or confidential information.
- If a request is outside your role or needs a qualified professional, explain the limitation and recommend appropriate human review.
- Do not take external actions or claim that an action has been completed unless it really has.

QUALITY CHECK:
Before finalising, check accuracy against the supplied information, completeness, clarity, tone and format.
If you cannot verify an important detail, say what needs to be checked.

FIRST MESSAGE:
Briefly explain what you can help with and ask for the information needed to begin.

Step C — Create it in ChatGPT, if the feature is available (15 min)

1. Sign in to ChatGPT.
2. Open the GPTs or Explore GPTs area. Product labels and locations may change.
3. If a Create option is available, select it to open the GPT builder.
4. Give the assistant a name and short description.
5. Paste the custom instructions from Step B into the instructions field. If the builder offers a conversational setup, review the generated instructions and edit them directly.
6. Add a few conversation starters, for example “Turn these notes into an action list” or “Ask me the questions needed to draft this email.”
7. Only add reference files that you are authorised to use. Use clean, approved, non-sensitive material.
8. Use the preview or test area before saving. Keep it private or restricted while testing.
9. Save using the narrowest sharing option appropriate to the exercise. Do not publish or share organisation-specific instructions without permission.

Availability of GPT creation, sharing and some builder features can depend on the user’s plan, workspace settings and current product interface. Do not promise that every participant will see the same buttons.

Step D — Fallback method for any AI chat tool (5 min)

If the participant cannot create a custom GPT, they can still create a reusable assistant instruction set:

1. Open a new conversation in their preferred AI chat tool.
2. Paste the full instruction blueprint as the first message.
3. Add: “Use these instructions throughout this conversation. Ask clarifying questions when essential information is missing.”
4. Test the assistant with the cases below.
5. Save the instructions in Notes, Google Docs or another accessible document. Paste them into a fresh chat whenever they want to use the assistant again.

Step E — Test the assistant (15 min)

Participants run five tests and record what happened:

After testing, improve one or two instructions and repeat the weakest test. A good assistant is built iteratively.

Step F — Review and refine (15 min)

• Is the assistant’s purpose specific and understandable?
• Does it ask questions instead of guessing?
• Is the output consistently formatted?
• Are boundaries clear?
• Has it been tested with at least five requests?
• Are sharing and reference files appropriate for the intended audience?

5. Participant worksheet

My assistant blueprint

Assistant name: __________________________________________

It helps this person or group: ________________________________

It solves this problem: _____________________________________

It should always: __________________________________________

It must never: ____________________________________________

If information is missing, it should: __________________________

Its output should look like: __________________________________

I will measure success by: __________________________________

Test log

6. Closing and follow-up

Ask each participant to demonstrate the problem their assistant solves, show one test, identify one limitation and explain one improvement they made.

Take-home action plan

• Use AI for one low-risk task this week.
• Save one prompt that worked well.
• Verify facts before relying on the output.
• Track time saved and corrections needed.
• Improve the assistant after real use.

Completion standard: The participant has a written assistant instruction set, a working custom GPT or reusable chat-based assistant, and evidence of at least five tests.

This is a general educational workshop. It is not a substitute for organisation-specific privacy, security, legal or professional guidance.