AI and automation projects fail when organisations bolt AI onto current processes instead of challenging whether those processes should exist in their current form.
Are you redesigning how your work operates, or just rebuilding what already exists with better tools?
There's a concept in urban planning called paving the cow path. You skip the design work, watch where people naturally walk, and lay tarmac on top of it. Efficient in the short term. Permanently encodes every inefficiency into your infrastructure.Most automation programs are doing exactly this.
When we walked our advisory board through a framework for AI and automation transformation, the feedback was direct. "Not impactful enough." "Not enough art of the possible."
These aren't people behind the curve. They're running automation programs at scale. And their frustration wasn't with the technology. It was with the ambition of the thinking around it. Too many organisations take existing processes and rebuild them with better tools. That's not transformation. That's renovation.
One executive framed it simply.
First, the Dewey Decimal System: you did the research yourself. Then Google found the source for you. Now ChatGPT gives you the answer. You're not hunting or evaluating anymore. You're questioning, directing, verifying.
Each shift didn't just change the tool. It changed the nature of the work itself. Bain's research on AI transformation backs this up: the companies seeing real ROI are those reimagining how work operates with AI embedded from the ground up, not layered on top of existing workflows.
Even in organisations with clear process ownership, those owners can still have blind spots across end-to-end processes. Multiple people own multiple slices, but how those slices connect into something that serves the actual business outcome? Nobody's explicit responsibility.
Which means when you try to reimagine the process, you don't have one conversation. You have twelve. With twelve people protecting twelve pieces of turf. That's not a technology gap. It's a structural and governance gap no automation tool can paper over.
The most sophisticated operators in the room had landed on the same sequencing: challenge the process logic and ask whether it should exist in its current form at all, before spending a dollar making it faster.
The question to ask before any automation initiative: if we were designing this from scratch today, with everything AI can now do, would it look anything like this?
If the answer is no, you have your brief.
Process reimagination isn't a technology project. It's a strategy project that technology then executes. It needs executive sponsorship, not just IT ownership. And it means your automation and AI partners need to bring this conversation. If the only thing on offer is a framework for digitising what already exists, that's implementation, not advisory.
Organisations that stop at implementation risk doing the wrong things at scale. Efficiently.
The technology has never been more capable. The question is whether you're pointing it at the right problem. If your organisation is ready to think differently about what's possible, let's talk.
