90% of companies have invested in AI. Fewer than 40% are seeing any measurable results.

That stat from McKinsey's Global Institute report, published November 2025 was the opening note at the European Chatbot and Conversational AI Summit in Edinburgh. And it set the tone for everything that followed.

Sooz Young attended as a moderator and panellist at the sixth edition of this summit, held at the University of Edinburgh from 17–19 March 2026. The theme was Harnessing Agentic AI. The thread running through every session, every room, every conversation was this: human centric.

What made this different from most AI events? These people weren't performing for a sales audience. They were talking to each other, honestly, about what actually works and what doesn't.

---

FIVE STANDOUT MOMENTS FROM OTHER SESSIONS

Omotenashi and what AI still can't replicate. Daniel Orenes Fernandez (Uber) opened with Japan's net promoter score: negative 52. In a country where the standard of service is so high that merely meeting expectations doesn't move the needle, the lesson for AI design is that you cannot just think about whether the system works. You have to think about what the cultural and emotional expectations surrounding the experience are and what the person on the other end actually needs.

Booking.com's five principles and the closing line. Raquel Serrano (Booking.com) walked through their five principles for human-AI collaboration: seamless integration, AI for speed and humans for trust, learning with every interaction, intent-driven anticipation, and ethical transparent guardrails. Her closing line: "The best customer service experiences are not the ones with the most AI. They're the ones with the most alignment between people, purpose and technology."

Raiffeisen Bank and the organisational design problem. The Raiffeisen story (covered in Thread 4) carries a lesson that scales to any business, any size: the problem is almost never the tool. It's the structure around the tool.

The value of design. Ray-Allen Taylor (IdeasBay): "The common illusion is that AI replaces design. The reality is that AI makes design more critical than ever." AI can generate, iterate, produce. It cannot decide what good looks like. It cannot weigh trade-offs between efficiency and integrity. Human judgment about what to build and why becomes more important as AI gets faster and better at the building.

The growth mindset shift and AI Scotland. Christoph Eslinger (IdeasBay) on the mindset shift organisations need: stop asking "where can AI cut costs?" and start asking "where can AI help us do things we couldn't do before?" Cost-cutting thinking makes you leaner. Growth thinking makes you different. On 20 March the day after the summit the Scottish Government launched AI Scotland, a new programme to help Scottish SMEs implement AI with ethical and legal frameworks built in. More details to follow as Sooz researches further.

---

🎓 PRACTICAL AI PROMPTING FUNDAMENTALS COURSE

Real business examples. Prompts you can copy, adapt and reuse straight away.

👉 https://allmylink.me/aipromptingfundamentals

🔍 DIGITAL SETUP AND SYSTEMS HEALTH CHECK

Understand exactly where your business stands across digital foundations, automation, AI, cybersecurity and data maturity in around 3 minutes.

👉 https://technologycoachingonline.com/digital-health-check

---

🧠 FIND YOUR DIGITAL STRENGTHS — Take the Free Quiz

Discover your digital persona and get a personalised action plan in under 3 minutes.

👉 https://technologycoachingonline.com/quiz/

🎉 JOIN THE DIGITAL DOMINATORS COMMUNITY — Free Access

Toolkits, templates, a Social Media Prompt Calendar, and a community of business builders who get it.

👉 https://technologycoachingonline.com/digital-dominators/

If Episode 42 made the case for why prompting is the real skill behind better AI results, this episode is where we get into the practical detail.

Sooz Young walks through the four most common prompting mistakes she sees repeatedly in training sessions, workshops, and client work — and shares four specific frameworks from the Practical AI Prompting Fundamentals Course that you can start applying straight away.

These methods work across every AI platform — ChatGPT, Claude, Gemini, Copilot — because the underlying skill is the same regardless of where you're prompting.

---

THE 4 MOST COMMON PROMPTING MISTAKES

1. Being too vague — asking for a blog post or some ideas with no context, audience, goal or format. The AI fills the gaps with generic information and the output reflects that.

2. Expecting AI to know your business automatically — unless you provide the context, the system has no idea who you are, who you serve, or how you communicate. This is especially common with free tools that don't retain memory between sessions.

3. Stopping after the first answer — the first response is a starting point, not the final output. The most effective users refine, iterate, and add more context. They treat AI as a collaborator, not a vending machine.

4. Searching for one magic prompt — clever formulas shared online create the impression that prompting is about memorising special phrases. In reality, effective prompting is clarity, context and iteration, built around reliable methods you can return to.

---

THE 4 FRAMEWORKS COVERED IN THIS EPISODE

Act As Method

Give the AI a role before it starts the task. Instead of "write a social media post about AI tools," try "act as a social media marketing strategist and write a carousel script for Instagram aimed at small business owners who feel overwhelmed by AI." Role, task, audience, purpose — that's what changes the output.

Ask Questions Method

Instruct the AI to ask you questions before it does the work. Ask it five questions about your audience, your offer, and your call to action — then answer them. The output is built on your actual context, not its best guess. It mirrors how a good consultant works, and it reduces the back and forth significantly.

PTCF Framework

The four-part structure behind any strong prompt: Persona (who the AI should act as), Task (what you want it to do — use specific verbs), Context (who is this for, where will it be used, what's the goal), and Format (bullet points, word count, tone, table). Most weak prompts are missing at least one of these four things.

https://allmylink.me/aipromptingfundamentals

Paste in your existing content — blog posts, emails, transcripts — and ask the AI to review your tone of voice and writing style before it starts. The new output picks up your rhythm, vocabulary and structure, and feels like you rather than a generic draft. Combine this with setting up dedicated projects in your AI tool (one for the podcast, one for proposals, one for social media, etc.) so the context carries across every conversation.

---

RESOURCES MENTIONED

🎓 Practical AI Prompting Fundamentals Course

👉 https://allmylink.me/aipromptingfundamentals

🔍 Digital Setup and Systems Health Check

👉 https://technologycoachingonline.com/digital-health-check

---

🧠 TAKE THE FREE DIGITAL MINDSET QUIZ

👉 https://technologycoachingonline.com/quiz/

🎉 JOIN THE DIGITAL DOMINATORS COMMUNITY — FREE

👉 https://technologycoachingonline.com/digital-dominators/

ChatGPT vs Claude. Claude vs Gemini. Gemini vs everything else.

Every few weeks there's a new headline declaring one AI tool better than the rest and every few weeks, people switch platforms hoping this one will finally deliver.

But here's what five years of teaching businesses how to use AI has taught Sooz Young: the tool is almost never the problem.

In this solo episode, Sooz unpacks the real reason most people get poor results from AI tools and why switching platforms will never fix it.

The issue is prompting. And prompting is a skill that almost nobody has been taught.

Since partnering with Facebook in 2019 to deliver AI and automation training to 3,000 people across Scotland, one pattern has shown up again and again: people are given access to powerful AI systems with no instruction on how to communicate with them effectively.

So when the outputs disappoint, they blame the tool. Then they switch tools. And the cycle continues.

This episode breaks that cycle.

---

KEY TAKEAWAYS FROM THIS EPISODE

- The "AI tool wars" explained — and why the debate is missing the point entirely

- Why most people treat AI like a search engine (and why that limits every result they get)

- The data and IP risk of using free AI tools — and why investing in the right tool matters

- How to choose the right AI tool based on your existing business ecosystem

- What prompt engineering actually means in plain, everyday language — no coding required

- The PTCF framework: Purpose, Task, Context, Format — how to structure any prompt

- A real example from a live training session: how changing the prompt (not the tool) transformed someone's output completely

- Why AI outputs sound generic: it's not hallucination, it's missing context

- The phrase to add to the end of any prompt: "ask me questions until you are 95–100% sure of the task and output"

- The "top 1% in my industry" prompt technique for more strategic, elevated results

- What you put in, you get out — the core principle behind every great AI interaction

- Why prompting is fast becoming one of the most essential skills in any job or business

---

RESOURCES MENTIONED

🎓 AI Prompting Fundamentals Course — Technology Coaching Lab

👉 https://allmylink.me/aipromptingfundamentals

📅 Next episode: Common prompting mistakes and how professionals structure their AI conversations

---

🔍 DIGITAL SETUP & SYSTEMS HEALTH CHECK

Identify the gaps in your digital infrastructure before they start costing you.

👉 ⁠https://technologycoachingonline.com/digital-health-check⁠

🧠 TAKE THE FREE DIGITAL MINDSET QUIZ

Find out where your digital strengths are — and where to focus next.

👉 https://technologycoachingonline.com/quiz/

🎉 JOIN THE DIGITAL DOMINATORS COMMUNITY — FREE

Toolkits, templates, a Social Media Prompt Calendar, early event access, and community.

👉 https://technologycoachingonline.com/digital-dominators/