Organizational lightness begins before the layoff.
The first visible sign of AI transformation is often a new tool. The deeper change appears later: fewer handoffs, shorter production cycles, wider spans of control and less tolerance for roles built mainly around collecting, formatting and forwarding information. An organization can become lighter long before its headcount formally changes.
This is why the current debate cannot be reduced to whether AI will “take jobs.” AI is changing the unit of work itself. Research, reporting, drafting, variation and coordination are being compressed into systems that one person can direct. The organizational chart may look the same for a while, but the logic that justified many of its layers is already under pressure.
The Diary of a CEO reveals both sides of the transition.
On The Diary of a CEO, former Google X executive Mo Gawdat forecasts that substantial job losses could arrive in selected sectors by 2027 or 2028. Yoshua Bengio similarly warns that jobs are already being replaced while power concentrates around the people who control advanced AI. These are warnings and forecasts—not settled labour statistics—but they reveal the urgency now shaping executive decisions.
The more instructive signal may come from the media organization around the show itself. In a separate interview, Flight Story co-founder Georgie Holt described an internal AI challenge that produced 48 tools and a self-reported 62,000 hours of efficiency, with the resulting capacity intended for reinvestment. The important point is not the headline number. It is that AI did not enter as one isolated application. It became a reason to redesign work across departments.
Organizational lightness begins before the layoff.
What disappears first is the work between the work.
AI does not need to replace an entire profession to reshape an organization. It only needs to absorb enough of the connective labour around that profession: assembling the weekly report, translating one format into another, preparing the first draft, monitoring signals, coordinating revisions and producing the variations required by each channel.
When these activities shrink, specialist teams can operate with fewer intermediaries. Managers can supervise broader systems. Small groups can produce at the scale of larger departments. This is organizational lightness: less friction between intention and execution. But it creates a new question. If the layers that once carried information, context and challenge disappear, where will those functions live next?