As a CEO, there are two ways your board can lose confidence in your artificial intelligence (AI) strategy. The first is watching a rushed roll-out fail in public. The second is watching nothing happen at all, quarter after quarter, while competitors move ahead. Leaders treat these as opposite failures: one of nerve, the other of discipline. But they are the same failure.Getting a motor vehicle fitness certificate used to mean an annual visit to a torture chamber. The nonsensical steps, inexplicable wait, and time lost translated into an activity everyone wished to avoid. It was a classic bottleneck in the process of being a Jamaican driver.But who could blame the workers at the facility? Diligent as they were, they were trapped in a system they had no power to change.Instead, the recently implemented changes show that the real solutions lay elsewhere, somewhere between the executive suite and their IT department. Decades of poor decisions by leadership, not the employees, kept common-sense changes from happening.This is a classic case of cause and effect being separated in location and time. In such cases, the department that breaks isn’t the department that failed. The quarter you see the failure isn’t in the same quarter in which the mistake was made.That’s why neither the rushed CEO nor the frozen CEO can see the shared cause. The consequences of skipping or delaying a process diagnosis show up later as a failed AI pilot, a stalled roll-out, or a board running out of patience. Each episode looks like an unrelated, isolated event.This same failure shows up in companies large and small, global and local. It’s not a resource problem or a development issue.For example, Starbucks recently scrapped an AI inventory tool after only nine months. Adidas automated one of its documentation processes and stripped out the human verification step that had been part of it – a change that upset stakeholders.These weren’t companies lacking money, talent, or AI enthusiasm.THE MISSING DIAGNOSISThese organisations possessed AI capability. They failed to diagnose the process itself before the tool was selected. This was an understandable gap, given the pressure to move quickly.In today’s fast-moving world, most companies are eager to seize AI opportunities, partly because stakeholders are demanding they keep up. When a new technology or automation shows up, they simply put it in place and measure the results.However, unless the change is as straightforward as installing a printer, one critical step gets skipped – a step that’s essential, given how complex even small companies’ processes are. There’s no step where the existing process is mapped, tested for bottlenecks and redesigned before automation is even considered.Both CEOs mentioned earlier would benefit from that same step. Instead of ‘move faster’ or ‘move slower’, they should insert the diagnostic step they aren’t used to taking.If the concept seems obvious, it’s worth asking why most companies have abandoned process management – a discipline built in the 1990s. Back then, people widely accepted managing an organisation’s core and supporting processes as a discipline. But over decades, leadership teams have let it lapse.Several years ago, a prospective client of my firm wanted to implement robotic process automation. The bad news I shared was that without process management, it would be a disaster. Today, the pressure to achieve AI-driven improvements has skyrocketed, but the need for process management upon initiation hasn’t changed. It’s simply been ignored.Today’s executives put too much faith in their own eyes. Some have experienced tremendous personal productivity gains with AI. Others have tried LLMs and found them lacking, deterred by the overconfident tone and the hallucinations.But these micro-experiences are misleading. The raw truth is that they don’t scale. They have little to do with enterprise-wide change that every organisation must consider.Instead, organisations need two elements. The first is process management capabilities across the entire organisation. The second is long-term strategic planning so that they can plan for decades of automation projects, not just a few months.It’s no fault of theirs – the pace of automation has simply outrun old habits.The fix isn’t a faster roll-out or a longer pause. It’s a mapped process, tested for its bottlenecks, before a single tool is chosen. And backed by a long-term road map of strategically selected process changes.Boards don’t reward the leader who moves first, or the one who moves last. They reward the one who knew where to look before moving at all.Francis Wade is the author of ‘Perfect Time-Based Productivity’, a keynote speaker and a management consultant. To search his prior columns on productivity, strategy, engagement and business processes, send email to columns@fwconsulting.com.