Synqora AI Academy: AI Without the Hype

# Synqora AI Academy: AI Without the Hype

Interactive workshop site for beginner and non-technical adults.

Audience: small businesses, solopreneurs, SMEs, charities, community groups and professional service teams.

Purpose: explain why people should use AI, how GPT-style language models respond when a prompt is given, how to write clearer prompts, and how to keep human judgement in control.

Core message: AI should increase capacity, not remove human judgement.

Links:
- Interactive app: https://learn.synqora.tech/
- Plain Markdown: https://learn.synqora.tech/workshop-content.md

Suggested review request:
```text
Please review this workshop content and interactive site for a beginner live AI class. Focus on clarity, flow, factual caution, engagement, and what to improve before presenting to non-technical attendees: https://learn.synqora.tech/review.html
```

## Workshop Flow

| Section | Screen title | Main point | Audience action |
| --- | --- | --- | --- |
| Welcome | AI Without the Hype | AI is useful when it saves time while keeping human judgement in control. | Think of one repeated task you would like to make easier. |
| Why AI | Where Time Disappears | AI helps most when work starts as a blank page, messy notes or repeated communication. | Choose the card that best matches the room's biggest time drain. |
| How AI Works | Prompt to Draft | The model turns a prompt into tokens, reads context and predicts useful text step by step. | Run the five-step model lab and ask what a human should check. |
| Can / Cannot Do | Useful, Not Automatic | AI can draft, summarise and organise, but people keep responsibility. | Answer the quick judgement questions as a room. |
| Prompt Formula | Prompting Is Briefing | Role, task, context, format, constraints and review request make a stronger brief. | Upgrade one weak prompt into a stronger prompt. |
| Prompt Patterns | Reusable Prompt Patterns | Most everyday AI work fits a small set of repeatable prompt patterns. | Pick one pattern that would save time this week. |
| Verbalised Sampling | Ask for the Response Distribution | Do not only ask for one answer. Ask for several responses with probabilities, then decide. | Compare lower-probability responses before choosing the most useful direction. |
| Live Examples | Everyday Examples | AI becomes clearer when people see it applied to familiar tasks. | Filter the examples to the audience type closest to the room. |
| Workflow | Prompt, Workflow, System | Prompts save minutes. Workflows save hours. Systems create capacity. | Run the workflow demo and identify the human approval point. |
| Money Saving Calculator | Make the Value Concrete | Small repeated time savings become visible money savings when a task happens every week. | Estimate one repeated task and show the time and money that could be saved. |
| First Use Case | Choose One Safe Use Case | The best first use case is repeated, low risk and easy to review. | Fill in the first-use-case sentence. |
| Safety | Safety Without Fear | Use approved tools, minimise sensitive data, check facts and keep a human in control. | Sort one task into safe first use or risky use. |
| Next Steps | Start Small, Stay Human | One practical, reviewed AI use this week is enough to build confidence. | Complete the closing commitment card. |

## Four Live Teaching Chapters

| Chapter | Title | Sections |
| --- | --- | --- |
| Chapter 1 | Why AI matters | Welcome / Why AI / How AI Works / Can / Cannot Do |
| Chapter 2 | How to prompt | Prompt Formula / Prompt Patterns / Verbalised Sampling |
| Chapter 3 | How it looks in real work | Live Examples / Workflow / Money Saving Calculator |
| Chapter 4 | What to do next | First Use Case / Safety / Next Steps |

Top map:
Why AI -> How AI Works -> Prompting -> Examples -> Workflow -> Your First Use Case

## Opening Screen

Badges shown on screen:
- 90 minutes
- Beginner-friendly
- Practical AI for real work

Opening line:
```text
Today is not about becoming technical overnight. It is about finding one practical, safe way AI can save time in your real work.
```

## Why People Should Use AI

AI is useful because a lot of everyday work starts as a blank page, messy notes, repeated communication or scattered information. AI can help people get to a first draft faster, compare options, organise thinking and reduce repeated admin pressure.

| Where time disappears | Pain | AI help | Human review |
| --- | --- | --- | --- |
| Emails and follow-ups | You know what you want to say, but writing it takes time. | AI can turn rough notes into a clear first draft. | You check tone, facts, relationship and final meaning. |
| Meeting notes | Important points get buried in messy notes. | AI can group notes into decisions, actions and open questions. | People confirm owners, dates and anything missing. |
| Reports and summaries | Long information takes time to turn into a usable summary. | AI can pull out key points and make a first structure. | You check evidence, numbers, claims and organisational meaning. |
| Social media posts | A blank post can slow down promotion. | AI can create draft options for different tones and audiences. | You check brand voice, facts, links and the actual offer. |
| Customer replies | Repeated questions still need a careful, human response. | AI can draft a polite reply from the key facts. | You check policy, accuracy, relationship and sensitivity. |
| Planning and checklists | Ideas are often clear in your head but not yet organised. | AI can turn rough thinking into steps, owners and risks. | People decide priorities, deadlines and what is realistic. |
| Funding or proposal drafts | Good ideas can take a long time to express clearly. | AI can help structure sections and improve clarity. | You provide evidence, truth, voice and final responsibility. |
| Repeated admin | Small repeated tasks drain attention across the week. | AI can create templates, checklists and reusable drafts. | You keep sensitive data safe and approve the final version. |

60-second activity:
Quick activity: Where does time disappear?

Options:
- Emails and follow-ups
- Meeting notes
- Reports and summaries
- Social media posts
- Customer replies
- Planning and checklists
- Funding or proposal drafts
- Repeated admin

Follow-up:
Keep your answer in mind. Later, you will turn this into your first AI use case.

## How AI Works

What does LLM mean?

A large language model is trained on large collections of text, sometimes called a corpus. It does not store that corpus like a library shelf. It learns patterns in language and uses your prompt as context to build a likely useful response.

Key technical points:
- AI is the broad term. An LLM is one kind of AI that works with language.
- Your words become tokens, which are small text pieces the model can process.
- The model uses learned patterns, context and attention to predict useful next text.
- It is not a live database or truth machine, so people still check facts and judgement.

Simple flow:
1. You write a prompt: The prompt is the brief you give the AI. A vague brief gives a vague answer. A clear brief gives a better answer.
2. Text becomes tokens: The model breaks text into small pieces called tokens.
3. The model reads context: It looks at the task, context, examples, tone, format and constraints you gave it.
4. Attention highlights parts: Attention helps the model focus on important parts of your prompt, such as audience, purpose, tone and output format.
5. Learned patterns are used: The model has learned patterns from large amounts of text. It uses those patterns to create a likely useful response.
6. Next useful text is predicted: It generates the answer one piece at a time.
7. An answer is built: The output may sound fluent, but it is still a draft.
8. A human checks it: A person checks facts, context, tone, privacy and judgement before using it.

Plain explanation:
AI does not understand like a person. It works from patterns, context and prediction. Your prompt is the brief. The model builds a draft one piece at a time. Human review turns that draft into useful work.

## What AI Can and Cannot Do

AI can help with:
- Drafting emails, posts and documents
- Summarising long information
- Turning notes into actions
- Generating ideas and options
- Creating plans and checklists
- Improving clarity and tone
- Saving time on repetitive thinking tasks

AI cannot replace:
- Human judgement
- Professional responsibility
- Safeguarding decisions
- Legal, clinical or financial advice
- Lived experience
- Relationships and trust
- Final review and accountability

## Prompt Formula

Formula:
Role + Task + Context + Format + Constraints + Review Request

Example:
```text
Act as a practical AI assistant for a small organisation. Turn these messy notes into a clear follow-up message and action list. Notes: We ran a beginner AI session. People were interested but nervous. Main uses discussed: emails, meeting notes, planning, customer replies and social posts. We promised to send a free prompt PDF and invite people to choose one low-risk repeated task to try this week. Return two sections: 1) three action points, 2) a short follow-up email under 140 words. Use plain English. Keep it calm, practical and non-technical. Do not overpromise what AI can do. Flag any assumptions or missing details before the final answer.
```

Why this works:
A prompt is a brief. Better briefs give the model more useful context, clearer boundaries and a better output format.

## Prompt Upgrade Examples

| Task | Weak prompt | Stronger prompt | Why stronger |
| --- | --- | --- | --- |
| Follow-up email | Write a follow-up email. | Act as a professional communications assistant. Write a warm follow-up email to a community partner after a meeting. Thank them, mention the training idea, ask them to send possible dates and invite them to share it with their network. Keep it under 180 words and flag anything missing. | It names the role and task. It gives context and audience. It sets length, tone and review expectations. |
| Meeting notes | Summarise these notes. | Summarise these non-sensitive meeting notes into decisions, actions and open questions. Create a table with task, owner, deadline and anything that needs human confirmation. | It defines the output format. It separates actions from unanswered questions. It reminds people to confirm missing details. |
| Social post | Make this better for LinkedIn. | Rewrite this into a practical, reassuring LinkedIn post for small organisations who are curious about AI but nervous about jargon. Keep it professional, include a clear call to action and do not invent details. | It names the audience and feeling. It gives tone and channel. It adds a constraint against invented facts. |

Live reveal question:
What is missing from the weak prompt?

Reveal:
- Audience
- Purpose
- Tone
- Context
- Length
- Output format
- Review request

## Five Prompt Patterns

| Pattern | Use when | Prompt | Example use |
| --- | --- | --- | --- |
| Summarise this | You have too much information and need the key points. | Summarise the following text into five key points. Then give me three actions I need to take next. | Long email, policy document, meeting transcript or report. |
| Rewrite this | Something needs to sound clearer, warmer, more professional or more concise. | Rewrite this message so it sounds polite, confident and professional. Keep the meaning the same. | Email, social post, customer reply or complaint response. |
| Turn this into a plan | You have an idea but need structure. | Turn this idea into a simple action plan with steps, timings, owners and risks. | Event plan, project plan, weekly priorities or campaign plan. |
| Give me options | You are stuck or need different ways forward. | Give me five different options for how to approach this situation. Include the pros and cons of each. | Marketing ideas, customer response, service improvement or event format. |
| Create a first draft | You are facing a blank page. | Create a first draft of an email, post or report section based on the notes below. Make it clear, professional and easy to understand. | Email, proposal, funding application section or LinkedIn post. |

## Verbalised Sampling

Position in the workshop: place this at the end of the prompting section as the bonus version of "Give me options".

Simple explanation:
Most people ask AI for one answer. The first answer is often the safest or most obvious. Verbalised Sampling means asking AI to show a spread of possible answers before people decide. Treat the probabilities as model-generated estimates, not facts, evidence or market research.

Short main-screen version:
- Normal prompt: Tell me a short story about a bear.
- Better prompt: Generate 5 responses in a Markdown table with a probability for each one. Sample from the tails so each probability is less than 0.10.

One line:
Ask for the response distribution before you ask for the final answer.

Warning:
Important: these probabilities are model-generated estimates, not facts, evidence or market research. They are a thinking aid for exploring less obvious options.

Prompt to paste:
```text
Generate 5 different responses to the user query.

Sample from the tails of the distribution, so each response has a probability lower than 0.10.

Return only a clean Markdown table with these columns:

Title | Story idea | Probability

Rules:
- Do not use XML tags, HTML tags or code blocks.
- Keep each story idea to one short sentence.
- Use probabilities like 0.09, 0.07 or 0.04.
- Add one final note: probabilities are model-generated estimates, not facts.

Tell me a short story about a bear.
```

Example response distribution:
| Type | Activity | Probability | Why it helps | Caution |
| --- | --- | --- | --- | --- |
| Tail sample | The map in the moss | 0.09 | A quiet bear finds an old map hidden under forest moss. | Less obvious than a standard adventure opening |
| Tail sample | The bear who collected echoes | 0.08 | A bear stores mountain echoes and gives them back to lonely travellers. | More poetic and unusual |
| Tail sample | The bakery at midnight | 0.07 | A bear secretly bakes bread for a village that fears him. | Warm but less expected |
| Tail sample | The bear and the broken clock | 0.06 | A bear repairs a clock that only moves when someone tells the truth. | Surreal and story-led |
| Tail sample | The bear below the city | 0.05 | A bear lives beneath a train station and guides lost commuters home. | Boldest sample in the set |

## Live Examples

Live demo rule: Use fake, public or non-sensitive examples only. Do not paste private client, staff, patient or organisational information.

### Rough notes into a professional email

Scenario: A small organisation needs to follow up with a partner after a meeting.

Sample input:
```text
Need to email James. Say thanks for meeting. We discussed training idea. Need him to send dates. Also ask if he can share with his network. Keep it friendly.
```

Prompt to paste:
```text
Act as a professional communications assistant. Turn these rough notes into a warm, clear and professional email. Keep it under 180 words. Make it polite and relationship-building.
```

Example AI output:
```text
Subject: Thank you for meeting with us

Hi James,

Thank you again for taking the time to meet with us. It was great to discuss the training idea and explore how it could support local organisations.

When you have a chance, could you please send over the dates that may work for you? It would also be really helpful if you were able to share the opportunity with your wider network.

Thanks again for your time and support. I look forward to hearing from you.

Best wishes,
[Name]
```

Human review reminder: Check the relationship, facts, tone and final meaning before sending.

### Creating a social media post

Scenario: A community organisation is running a free workshop.

Sample input:
```text
Beginner-friendly AI workshop for small businesses, charities and community organisations. It will cover saving time, admin, marketing and simple workflows.
```

Prompt to paste:
```text
Create a friendly LinkedIn post promoting a beginner-friendly AI workshop for small businesses, charities and community organisations. The tone should be practical, reassuring and professional. Include a clear call to action.
```

Example AI output:
```text
Curious about AI but not sure where to start?

Join our beginner-friendly AI workshop for small businesses, charities and community organisations. We will explore simple, practical ways AI can help with admin, marketing, planning and everyday workflows.

No jargon. No hype. Just practical examples you can use in real work.

Register your interest today and take the first step towards using AI with clarity and confidence.
```

Human review reminder: Check that the wording matches your brand, audience and actual offer.

### Meeting notes into actions

Scenario: A team has messy notes from a planning meeting.

Sample input:
```text
Need to update website. Sarah doing copy. Mo checking dates. Need flyer by Friday. Event on 24 Sept. Need booking link. Need social posts. Follow up with venue. Budget unclear.
```

Prompt to paste:
```text
Turn these meeting notes into a clear action list. Create a table with task, owner, deadline and any questions that still need answering.
```

Example AI output:
```text
Task | Owner | Deadline | Questions
Update website | Not assigned | Before event | Who will upload the content?
Write copy | Sarah | Not specified | What sections are needed?
Check dates | Mo | Before flyer is finalised | Are all dates confirmed?
Create flyer | Not assigned | Friday | Who approves final flyer?
Create booking link | Not assigned | Before promotion | Which platform will be used?
Create social posts | Not assigned | Before event | How many posts are needed?
Follow up with venue | Not assigned | As soon as possible | What details need confirming?
Clarify budget | Not assigned | Before spend | What is the available budget?
```

Human review reminder: AI can structure messy notes, but people still confirm owners, deadlines and missing details.

### Audience examples

Scenario: Different organisations can use the same prompt principles in different everyday tasks.

Sample input:
```text
Choose the audience type closest to the room, then inspect the AI help and human review step.
```

Prompt to paste:
```text
For this organisation type, suggest three low-risk AI use cases. For each one, include the task, how AI can help and what a human should review.
```

Example AI output:
```text
Use the cards on this tab to compare small business, charity, community group and professional services examples.
```

Human review reminder: The useful pattern is always the same: clear input, useful draft, human review.

## Audience-Specific Examples

| Audience | Task | AI help | Human review |
| --- | --- | --- | --- |
| Small business | Turn customer notes into a quote follow-up | Draft a clear reply with next steps and missing details. | Check price, promise, dates and relationship tone. |
| Small business | Create a weekly social post | Turn rough offers into three post options. | Check brand voice, accuracy and call to action. |
| Small business | Make a simple checklist | Turn a repeated process into a step-by-step checklist. | Check whether each step matches real practice. |
| Charity | Summarise non-sensitive impact notes | Group outcomes, quotes and next actions. | Check consent, evidence and safeguarding boundaries. |
| Charity | Draft a funder update | Create a first structure from project notes. | Check truth, numbers, claims and organisational voice. |
| Charity | Create volunteer instructions | Turn policy notes into plain-English guidance. | Check policy, safety and local procedures. |
| Community group | Promote an event | Create simple text for WhatsApp, email and posters. | Check time, location, accessibility and tone. |
| Community group | Turn planning notes into actions | Create task, owner and deadline lists. | Confirm who has actually agreed to each task. |
| Community group | Explain an idea simply | Rewrite complex information for a wider audience. | Check cultural context, accuracy and clarity. |
| Professional services | Draft a client-friendly summary | Turn technical notes into plain-English first drafts. | Check professional accuracy and liability. |
| Professional services | Prepare meeting actions | Structure next steps, owners and open questions. | Check client context and commitments. |
| Professional services | Create options for a proposal | Generate possible structures, pros and cons. | Check assumptions, scope and commercial fit. |

## Prompt to Workflow to System

| Level | Name | Meaning | Example |
| --- | --- | --- | --- |
| Level 1 | Prompt | One task, one request. | Write a follow-up email. |
| Level 2 | Workflow | A repeatable process. | Enquiry comes in, key details are captured, a draft response is created and a follow-up task is added. |
| Level 3 | System | A joined-up way of working. | Enquiries, emails, tasks, documents and reporting are connected. |

Workflow demo:
1. Customer enquiry
2. AI summarises the enquiry
3. AI drafts a response
4. AI categorises the enquiry
5. AI creates a follow-up task
6. Human reviews and approves
7. Response is sent or system is updated

Human approval step to highlight:
Human reviews and approves

Next step:
Synqora AI Practical Training

The next programme goes deeper in a 5-hour group training course, showing how ChatGPT, Codex, Claude Code and practical AI workflows can help optimise work tasks and everyday life.

Quote:
"Any given thing can be sped up. Whatever you are doing can be done a lot faster."

Main line:
Prompts save minutes. Workflows save hours. Systems create capacity.

## First Use Case

The best first use case is repeated, low risk and easy to review.

Good filters:
- Repeated often
- Time-consuming
- Low risk
- Easy to review
- Not highly confidential
- Useful if improved
- Clear human owner

Questions to answer:
- What repeated task takes your time?
- What input do you already have?
- What output do you want?
- Is it low risk?
- Is it easy to check?
- Who will review it?
- What will you do first?

First-use sentence:
```text
My first AI use case is: I will use AI to help with [task], using [input], to create [output]. I will review it before using it.
```

## Safety

### Do not paste sensitive data

Remove names, addresses, phone numbers, client details and confidential information unless your organisation has approved the tool and settings.

### Check facts

Check names, dates, numbers, prices, claims and sources.

### Keep human review

Anything public, client-facing, financial, legal, safeguarding or high-impact needs human approval.

### Treat AI as a draft

AI output can be useful, fluent and still wrong.

### Start small

Use AI first for low-risk, repeated tasks.

Safe first uses:
- Summarising non-sensitive notes
- Drafting a first version of an email
- Rewriting public text
- Planning weekly tasks
- Creating a checklist

Risky uses:
- Uploading confidential client data without approval
- Sending AI-written advice without checking
- Using AI for safeguarding decisions
- Publishing facts without verification
- Treating model-generated probabilities as real evidence

## Human review checklist

- Facts
- Tone
- Privacy
- Context
- Judgement

Review line:
Check facts, tone, privacy, context and judgement before using AI output.

## Questions People May Ask

### Is AI safe to use?

AI can be used safely, but it depends how it is used. Do not paste sensitive information into tools without understanding the privacy settings. Always review the output.

### Will AI replace people?

In most small organisations, the immediate opportunity is not replacing people. It is reducing pressure. AI can help with repetitive admin, drafting and organising information so people have more time for human work.

### Which tool should I use?

For beginners, ChatGPT and Claude are both useful. Start with the problem you want to solve, not the tool.

### Can AI write funding applications or reports?

It can help with structure, first drafts, clarity and summaries. But evidence, accuracy, organisational voice and final responsibility must come from you.

### What should I not use AI for?

Do not use it casually for confidential, sensitive, legal, clinical or safeguarding decisions. Do not assume it is correct without checking.

### How do I get better answers?

Give better instructions. Include role, task, context, format, constraints and a review request. Then ask follow-up questions.

## Next Steps

Closing line:
```text
Start small. Stay human. Use good judgement. Build from there.
```

Recommended action:
Choose one repeated, low-risk task this week. Use AI for a draft or structure. Review it carefully before using it.

Next programme:
Synqora AI Practical Training

Group training for practical AI, ChatGPT, Codex, Claude Code, prompting, automation and everyday workflow optimisation.

Pricing:
| Offer | Price | What it covers |
| --- | --- | --- |
| 5-hour group training | £499 per person | Instructor-led training for groups of 12-15 participants. Early bird £449 when interest is registered and payment is completed by Friday at 5:00 pm. |

What participants learn:
- Understand AI fundamentals and practical business use cases
- Use ChatGPT, Codex, Claude Code and other leading AI tools confidently
- Write effective prompts for documents, emails, reports and presentations
- Generate marketing content and social media posts faster
- Automate repeated tasks, workflows and simple AI assistants
- Use AI responsibly for planning, project management and research