Nearly eight in ten companies report using gen AI—yet just as many report no significant bottom-line impact. WHY?

“It took changing the system to suit the tool rather than adapting the tool to suit the system, that unlocked extraordinary gains.”

EXECUTIVE SUMMARY

Most companies are investing in AI but failing to see a significant return. The problem isn't the technology; it's the approach. Simply bolting AI onto existing, outdated processes is like replacing a steam engine with an electric motor but keeping the inefficient factory layout. True transformation—and a massive competitive edge—comes from fundamentally redesigning work around AI's unique capabilities. This article explains why this shift is critical and introduces a proven four-principle framework for achieving real ROI by reimagining roles, proving value in sprints, and empowering your people through a safe, human-centric transformation.

THE AI ROI PARADOX

‘Nearly eight in ten companies report using gen AI—yet just as many report no significant bottom-line impact’. According to a report from McKinsey this June 2025.

Which is not surprising, and was largely foretold. By many.

The big issue is that, as happens with all new technologies, companies tend to try and bolt them on to existing systems. We’ve spent, in many cases, decades ‘digitising the past’ and now many are set on ‘AI’ing the past’.

This approach either never works or doesn't get you very far, because the process itself is the impediment. A hammer is great if you have nails to hit, but less good if you’re working with eggs. Introducing a new technology can be as dramatic a change as moving from nails to eggs.

The groundbreaking technologies enable you to do things you simply could not before, so the entire architecture of what you do can be fundamentally different. It’s only when you align all the moving parts, and redesign ‘the system’ do you see great returns.

A LESSON FROM THE AGE OF ELECTRICITY

The archetypal analogy is the steam factory at the dawn of the electricity age.

Steam powered factories essentially had one large engine powering a single large driveshaft which controlled winches and pulleys and formed the centrepiece of all activity. So swapping the technology powering the driveshaft from steam to electricity made very little difference. Work carried on as before. No-one really noticed.

So from the early 1880s when electricity started to become available as a power source, until the 1920s, the new technology had very little effect on productivity.

In the early 1920s though they started to redesign the factory itself. Instead of one central drive shift, work was split up (Adam Smith’s ‘Division of Labour’) into multiple workstations each with their own electric source of power.

And productivity exploded!

It took changing the system to suit the tool rather than adapting the tool to suit the system that unlocked extraordinary gains.

FROM ELECTRICITY TO AI: WHY THIS TIME IS FASTER

And today is exactly analogous to this. We are largely pre-redesign and getting exactly what could be expected.

Except in two ways. First, whilst electricity is a ‘General Purpose Technology’, meaning it has an effect across an entire economy, AI is a ‘GPT’ in a much more obvious way (the GPT in ChatGPT actually stands for ‘Generative Pre-trained Transformer’). Electricity was so novel it took a long time for its use to disseminate, but the pervasive consequences of AI are much easier to see. We can see where it can be leveraged very easily. This does not mean across the board adoption will be instantaneous but it does suggest that the timelines for adoption are probably much shorter. Maybe think a decade rather than forty years.

And secondly AI is something that certain industries (think Software, Marketing), especially startups, can leverage, at scale very very quickly. Hence the rise of the fast, agile, ultra-productive superteams we’ve discussed before.

So competition to all is going to come much faster. Yesterday I was at the CRETech London conference, and someone from JLL noted that they were already seeing ‘fee compression’, particularly in marketing. Customers are realising that suppliers have access to tools that should enable them to do more, much faster, and so are expecting lower costs.

Try competing in a market expecting different price points because of technology, when you are not using those technologies. It’s a very fast route to margin erosion, even bankruptcy.

So those 80% of companies have got to get their acts together and work out how to extract real ROI from AI.

We think we have the solution to that. With …

OUR FOUR GUIDING PRINCIPLES

Our methodology is designed to be both transformative and safe. It balances bold strategic goals with practical, human-centric execution.

The fundamental aim is to maximise your impact by enabling you to focus on where you add the most value. Automate routine tasks, augment human capabilities, and cultivate ‘AI Synergy’—where humans and machines together achieve outcomes greater than either could independently.

This is achieved through four core principles.

1. Reimagine the Role, Not Just the Task.

• The Principle: We don't just "AI the past." Our primary goal is to fundamentally redesign work by unbundling roles into their core tasks. We then analyse and rebundle the role around its highest-value, uniquely human functions—creativity, strategic thinking, and client relationships—while AI handles the rest. This creates new capacity and more fulfilling work.
• Why it Matters: This is how we move beyond simple efficiency gains to create a true strategic advantage and a more empowered workforce.

2. Prove Value in Sprints, Then Scale with Confidence.

• The Principle: Transformation starts with focused, evidence-based experiments. We don't bet the firm on an unproven idea. We use rapid micro-sprints (2-4 weeks) to test a new, AI-augmented workflow on a small scale.
• Why it Matters: We only scale what works. Every sprint must conclude with a clear "ROI Sketch" that demonstrates concrete value. This data-driven approach de-risks innovation and ensures we only invest in proven, next-generation workflows. This combines starting small with the discipline of proving value before scaling.

3. Empower the Person, Govern the Platform.

• The Principle: This is our core social contract. We empower your people with powerful tools and the autonomy to innovate, but we do so within a framework of strong governance. The human is always the expert-in-the-loop, accountable for the final output.
• Why it Matters: This builds trust. It tells your team that AI is a tool to augment them, not replace them. And it tells your leadership that we are managing risk, protecting IP, and ensuring security by governing the platforms, prompts, and data they use.

4. Capture & Compound the Learning.

• The Principle: A single success is a victory; a shared success is a compounding capability. The knowledge gained from every sprint—both successes and failures—is a valuable asset that must be captured and shared.
• Why it Matters: We build a living "Process & Prompt Library" that becomes your firm's central playbook for operational excellence. This creates a powerful flywheel effect, allowing the entire organisation to get smarter, faster, and more innovative with every cycle.

Bringing It All Together

• Principle 1: Strategic Vision
• Principle 2: Agile Process
• Principle 3: Human & Safety Ethos
• Principle 4: Organisational Learning Engine

Collectively, these principles form a coherent, resilient, and adaptive framework that transforms AI from mere technology deployment into strategic, human-centred competitive advantage.