Eight months ago, AI still felt like a collection of possibilities.

Since then, we've spoken to business leaders, technologists, lawyers, engineers and AI practitioners, all trying to answer the same question from different perspectives. Along the way, some predictions have become reality, others have changed entirely, and a handful of themes have surfaced time and time again.

In the final episode of Season 3, Steve Clarke and Alastair Jupp look back at those conversations. They revisit the ideas that have stood the test of time, reflect on what surprised them, and discuss where AI is heading next.

From data and governance to culture, judgment and business evolution, Hallucinations of Grandeur is a chance to pause, take stock and consider what the last eight months have taught us.

In this episode of QInsight, Steve Clarke and Alastair Jupp are joined by Jamie Plumridge, Channel Director at nLighten, to explore the infrastructure that sits beneath every AI model, chatbot, agent, and productivity tool.

As AI adoption accelerates, demand for computing power is rising faster than the infrastructure supporting it. The discussion examines what happens when technology starts colliding with physical limits, from power consumption and cooling requirements to the growing pressure on energy networks and data centre capacity.

Jamie explains why the cloud is far less abstract than most people imagine, how modern AI workloads are changing the economics of infrastructure, and why the industry is now considering solutions that would have sounded like science fiction only a few years ago. The conversation covers hyperscale data centres, liquid cooling, regional AI infrastructure, and the growing role of nuclear energy in supporting the next phase of AI growth.

The episode also explores what this means for businesses. Most organisations will never build their own AI infrastructure, but every organisation will depend on it. Understanding those dependencies is becoming just as important as understanding the tools themselves.

If AI feels limitless, this episode is a useful reminder that every prompt, model, and agent still depends on power, hardware, and the laws of physics.

Steve Clarke and Alastair Jupp welcome back Steve Gibson from Lumen Consulting to revisit a conversation they first had earlier in the year and explore just how much has changed since then.

The discussion follows AI's rapid evolution from tools that answer questions to systems that can carry out tasks, complete workflows, and increasingly operate as part of a business itself. Along the way, the team explore Microsoft's latest developments, the growing role of AI agents, and why keeping up with the pace of change is becoming a challenge in its own right.

But this episode is less about technology and more about what businesses choose to do with it.

Rather than focusing on headcount reduction and efficiency targets, the conversation examines a different opportunity for SMEs. AI gives smaller organisations access to skills, expertise, and capabilities that were previously difficult or impossible to afford. The real question is no longer how to do the same work faster, but what becomes possible when expertise becomes accessible on demand.

The team also discuss workflow design, governance, human oversight, and why understanding how your business actually operates may be more important than adopting the latest AI tool.

If AI is moving in dog years, this episode is about making sure your business keeps pace without losing sight of what matters.

Everyone wants AI. Nobody wants to talk about the data.

In this episode of QInsight, Steve Clarke and Alastair Jupp are joined by Ryan Pennett, VP of Business Development at QWare, to explore why so many AI projects struggle before they have even begun.

The problem is rarely the technology. It is the quality of the information underneath it. From duplicate records and outdated customer information to missing ownership and weak governance, the discussion looks at the reality of what happens when businesses point AI at data that has been neglected for years. The result is often expensive automation, unreliable outputs, and processes that move faster without becoming any better.

Ryan shares practical examples of the issues organisations encounter when preparing for AI, why data quality remains one of the least glamorous but most important parts of digital transformation, and how businesses can begin tackling the problem without turning it into a massive programme of work. The team also explore AI consumption costs, the risks of automating bad processes, and why governance should be owned by the business, not left to IT alone.

If AI is the destination, this episode argues that most organisations still need to spend time fixing the runway before attempting take-off.

In the second episode of Season 3, Steve Clarke and Alastair Jupp are joined by Chris Roberts, Director at E&F Solutions, to explore why successful CRM and AI projects start long before anyone talks about software.

Chris describes himself as a translator, someone who bridges the gap between operational teams and technology. That idea runs through the entire episode. Projects rarely fail because the software isn't capable. They fail because nobody has taken the time to understand how the business really works.

The discussion moves from CRM implementation to AI, data quality, governance and regulation, but always comes back to the same principle. Start with the process. Talk to the people doing the work. Then decide how technology can support them, rather than expecting technology to fix a broken process.

Whether the project is CRM or AI, Chris makes the same point throughout. Understand the process first. The technology comes afterwards.

In the first episode of season 3 QInsight, Steve Clarke and Alastair Jupp are joined by Sandra Koutzenko, Director of Transformation at 365 Talent Portal, to discuss what AI adoption looks like when you move beyond the headlines and into the reality of running a business.

The conversation explores why AI feels overwhelming, not because the technology is difficult to access, but because the volume of change is relentless. New tools appear every week, vendors are all telling a different story, and businesses are being asked to make decisions in an environment that refuses to stand still.

Sandra shares her perspective on choosing AI platforms, the risks of becoming locked into a single approach, and why Microsoft Copilot is emerging as a practical foundation for organisations looking to build long-term capability.

The discussion also covers AI fatigue, employee resistance, governance, AI literacy, and the arrival of Cowork, Microsoft's latest step towards genuinely delegated digital work.

Rather than focusing on which model is winning this month's race, the episode looks at how businesses can build the confidence and flexibility to keep moving as the technology evolves.

For the season finale, Steve Clarke and Alastair Jupp are joined by Dr David Panni, bringing over 25 years of engineering and real-world AI experience to the conversation.

This episode tackles one of the most important ideas in AI right now. These tools are powerful, but they do not remove responsibility. They support the work, they accelerate it, but ultimately, you are still accountable for the outcome.

The discussion explores how AI is changing the skills businesses value, why asking better questions is becoming more important than knowing the answers, and how organisations should think about capacity, productivity, and long-term competitiveness.

It is a fitting close to the season. Practical, honest, and focused on what AI actually means for businesses today.

Season 2 - Episode 5: Goverance & Guardrails

In this episode of QInsight, hosts Steve Clarke and Alastair Jupp are joined by Kim Simmonds, CEO of Law 365 and Cloud Contracts 365, to explore what responsible AI adoption really requires as experimentation turns into operational reality.

Kim draws parallels between today’s AI uncertainty and the early days of GDPR, when organisations knew change was coming but lacked practical clarity. The conversation unpacks the legal and governance questions businesses should be asking, from data protection and confidentiality to intellectual property and liability.

They examine the rise of shadow AI, the responsibility that sits at board level, and why governance should be a framework that enables progress rather than slows it down. The episode also tackles the risks of overreliance on AI output, the importance of keeping humans in the loop, and what “reasonable measures” look like in practice.

It’s a grounded discussion about building the right structures so businesses can move forward confidently, protecting trust, value, and reputation while embracing AI as a tool rather than treating it as an answer.

In this episode of QInsight, hosts Steve Clarke and Alastair Jupp are joined by Vincent Sider, internationally recognised AI strategist and executive advisor, to explore how AI is getting smaller, cheaper, and closer to home.

Vincent challenges the assumption that serious AI requires vast cloud platforms and enterprise-scale budgets. Through practical examples, he explains how powerful models can now run locally on modest hardware, and what this shift means for cost, performance, and control.

The conversation digs into local inference, open-source models, and the rise of on-prem AI, particularly for organisations concerned about data sovereignty, governance, and unpredictable cloud costs. They discuss when it makes sense to use platforms like OpenAI or Copilot, and when running AI locally becomes the more sensible option.

The episode also looks ahead to agentic AI, where specialised agents handle discrete tasks across sales, marketing, and operations, and why controlling cost, evaluation, and risk becomes critical as AI works continuously in the background.

It’s a practical, demystifying discussion about right-sizing AI, lowering barriers to adoption, and helping organisations move forward with confidence rather than fear.

Season 2 - Episode 3 - AI, Everywhere, All at Once

In this episode of QInsight, we’re joined by Steve Gibson, founder of Lumen Consulting, to talk about our AI adoption journey so far.

The conversation reflects the reality that many organisations are experiencing as AI becomes more visible across tools, workflows, and leadership conversations.

We explore what happens when experimentation moves faster than understanding. Steve shares examples of early enthusiasm, false starts, and moments where progress felt busy rather than useful. We discuss why many AI stories sound impressive on the surface but fail to translate into day-to-day value, and how easily activity can be mistaken for advancement.

The discussion also looks at how expectations around AI have shifted over the past year. We talk about where tools like Copilot are genuinely helpful, where they still fall short, and why human judgement remains essential. Culture, mindset, and permission to experiment emerge as recurring themes, particularly when teams are trying to move beyond novelty and towards something more sustainable.

The episode offers a grounded perspective on learning in public, creating space for honest reflection, and why successful AI adoption depends far more on people and process than on technology alone.