Transforming the way organisations use AI to create value; Stop automating broken processes

There’s a growing disconnect in the AI conversation; personal productivity tools and scattered approaches are not meeting expectations, putting the technology first over the time to rethink. The outcome is scattered, fragmented outcomes not shifting the dial to meet board room expectations.

Many organisations are discovering that digitising dysfunction doesn’t make it better - it just makes it faster.

MIT Sloan recently drew a sharp distinction between automating tasks and redesigning work. Productivity gains only emerge when leaders reconstruct processes end-to-end, not when they apply automation to yesterday’s workflows.

Before applying any AI or automation, it’s worth asking the hard question:

Is the process itself worth saving? Below we work through an approach to ensuring that your investments match expectations.

1. Start with purpose and pathway

Every AI pilot should have a clear ‘why’ - a measurable business outcome - and a ‘how’ - a defined pathway to integrate into operations.

Ask four questions before the first line of code is written:

1. What decision or process will this improve?
2. How will success be measured (time, cost, quality, risk, customer outcome)?
3. Where are people compensating for poor design rather than creating impact?
4. If successful, what’s the first step to embed it into the business?

The most powerful thing a business can do to invest in AI is to truly understand the engineering of its core process flows. As one colleague likes to put it: “Don’t freeze today’s dysfunction in code.”

Automating a broken process at scale only embeds inefficiency deeper and makes it harder to fix later. If you can’t answer those questions, you’re not ready to start the pilot.

2. Start with value pools, not use-cases

It’s easy to be dazzled by catalogues of AI use-cases - chatbots, document summarisation, predictive analytics. But organisations that succeed, start from value pools, not use-cases.

Ask:

• Which end-to-end process has the biggest cost, risk, or customer impact?
• Where would a 10 percent improvement meaningfully shift our strategic outcomes?
• What insight or automation, if applied here, would change the economics of how we work?

When AI investment begins with business logic, it earns trust and momentum. When it starts with shiny tools, it becomes just another IT experiment.

3. Redesign the work, not just the workflow

True transformation isn’t about automating steps, it’s about rethinking roles, decisions, and accountability.

That means combining technology with a business-led view of how work should flow, who owns each decision, and how humans and machines complement each other.

Leaders should treat this as a workforce-design challenge: create roles that leverage human judgement, empathy, and creativity, while using AI for scale, consistency, and prediction.

This is where technology’s art of the possible becomes a leadership question:

How do we want work to feel and perform in our organisation five years from now?

4. Pilot with a path to scale

Most AI initiatives don’t fail because the models underperform, they fail because the business architecture isn’t ready.

When you pilot, design it as a microcosm of the future state, not a disconnected proof-of-concept.

Use it to validate flow, data, metrics, and change readiness and then scale horizontally.

The shift is from proof of concept to proof of value. That subtle difference changes how teams design, govern, and measure success.

Example: Re-imagining claims processing

An insurer wanted to “automate claims triage with AI”. Instead, they began by mapping the full process (critically - how it actually works rather than the theoretical ideal) from intake and verification to decision and payout. They found that 40 percent of delays came from incomplete information and rework between teams.

They redesigned the intake process to capture critical data upfront and simplified decision routing. Only then did they introduce AI classification to assist complex cases.

The results spoke for themselves: claim turnaround improved by 35 percent, exceptions fell, and staff spent more time resolving outcomes than chasing paperwork.

AI didn’t replace work - it reshaped it.

5. From talk to tangible value

Getting past the chat about AI means treating it not as a side project but as a business capability, anchored in process design, measurable outcomes, and leadership intent.

The technology is the easy part.

The hard part is deciding what deserves to be automated at all.

Framing the shift

As with every industrial transformation, progress depends less on tools and more on imagination, the willingness to redesign how work creates value.

AI offers extraordinary potential, but only if we stop automating broken processes and start building better ones.