We bolted AI onto the old org chart. Why does everything feel more chaotic?

You dropped AI into the roles and processes you already had, and you expected it to slot in cleanly. Instead the organisation feels busier and more chaotic, not less, and the reason stays out of reach. The dashboards say adoption is climbing. The team is using the tools. And yet the week feels louder, the handoffs feel messier, and the value you were promised has not landed anywhere you can point to on the P&L.

Here is the short answer, before the diagnosis. AI does not create value when you add it on top of the existing org chart. It creates value when you redesign the jobs, the roles, the decision rights and the processes around the work. The technology works. The approach, bolting it on, is what fails. You did not buy the wrong tool. You wired a new engine into an old gearbox and asked it to drive more smoothly.

Why does adding AI to the org make everything feel more chaotic, not less?

Because you changed the speed of the work without changing its shape. A modern AI tool can draft, summarise, analyse and propose in seconds. But if the role it sits inside still has the same handoffs, the same approval gates, and the same ambiguous ownership it had a year ago, every one of those seconds of output now has to be checked, reconciled and re-decided by a human downstream. You did not remove the friction. You poured more volume through the same narrow pipes.

This is the named phenomenon behind the feeling. Researchers at BetterUp Labs and the Stanford Social Media Lab (a research group studying how people behave with technology) call it "workslop": AI output that looks like finished work but lacks the substance to advance the task, so the burden of interpretation lands on a colleague. In a September 2025 survey of 1,150 US desk workers, 40% had received workslop in the previous month, each instance taking nearly two hours to resolve, an invisible cost of about $186 per employee each month, or over $9 million a year for a ten-thousand-person organisation. The harm is downstream, in the rework and the eroded trust between roles. That is precisely what chaos feels like from the inside: more motion, less progress, and nobody quite able to name where it leaks.

Is this a tools problem or an org-design problem?

It is an org-design problem wearing a tools costume. The numbers are unusually clear on this point. McKinsey's 2025 survey found that only 21% of organisations using generative AI have fundamentally redesigned at least some workflows, and that fundamental workflow redesign has the biggest effect on an organisation's ability to see EBIT impact (the profit a business actually keeps from operations) from its use of generative AI. The same study found more than 80% of respondents see no tangible effect on enterprise-level EBIT. Read those two findings together and the picture resolves: the firms feeling the chaos are the ones who deployed the tool and left the work intact.

Deloitte's 2026 State of AI in the Enterprise survey, covering 3,235 leaders across 24 countries, puts a sharper edge on the same gap. Only 30% of organisations are redesigning key processes around AI, while 37% use it at a surface level that leaves the underlying process unchanged. The talent picture is starker still: only 33% are redesigning roles and career paths, even though the leaders surveyed name insufficient worker skills as the single biggest barrier to weaving AI into how the work runs. Most organisations spend their attention and budget on the tooling, because that is the part with a vendor and an invoice. The role redesign has no purchase order. So it gets skipped. The bottleneck is no longer the technology. The bottleneck is the org that ran the technology.

There is a quieter cost too, and it sits in the people you most need. When roles stay unchanged, AI does not reduce the number of decisions a person makes; it multiplies the micro-decisions and handoffs they must adjudicate. Decision quality is finite. A study of more than 1,000 judicial parole rulings found favourable decisions ran near 65% at the start of a session and fell towards zero before a break, then reset to 65% after the judges had eaten. Judgement depletes with use. Load your best people with an unending stream of AI-generated fragments to verify, and you spend their judgement on reconciliation rather than on the calls only they can make.

What does redesigning jobs, roles and decision rights around AI actually look like?

The solution category is organisational redesign, and it is more concrete than it sounds. It is the deliberate work of deciding, task by task, which work belongs to a human, which to a machine, and which to the two of them together, and then redrawing the handoffs and the decision rights to match. The World Economic Forum's 2025 outlook frames the destination plainly: by 2030 employers expect only 33% of tasks to be performed mainly by humans, down from 47% today, with 34% mainly by technology and 33% done in human-machine collaboration. That middle third is the whole game. It does not design itself, and it will not survive being grafted onto roles built for a fully human workflow.

This is where I will name the deeper pattern, for the reader ready for it. We are living through the third great shift in how value gets made. Phase One, the Age of Effort: work hard, get a little more, linear growth. Phase Two, the Age of Scale: build once, sell to millions, exponential growth. Phase Three, the Age of Acceleration: output decoupled from human effort almost entirely, the phase AI unlocks. Most leaders meet Phase Three by trying to install new software on broken hardware: a powerful model running on an operating system, the org's roles and rhythms, that was written for a slower world. The coherence you are missing is not in the tool. It is in the fit between the tool and the way your people actually work. Build that fit, and the chaos resolves into capability.

So the path forward is sequential, and the order matters.

1. Map the work, not the tool. Trace one core process end to end and mark every handoff and every decision point. The friction lives in the seams, and the seams are where AI output currently piles into a backlog.
2. Re-sort the tasks. For each step, decide whether it belongs to a human, to the machine, or to the two together. Aim for that collaborative third deliberately rather than by accident.
3. Redraw the decision rights. Name who owns each call now that the inputs arrive faster. Faster drafting needs clearer ownership, or the speed just creates more to adjudicate.
4. Rebuild the roles around the new shape. Reshape the jobs to match the redesigned process. The WEF finds 85% of employers plan to prioritise upskilling and 59% of the workforce will need reskilling or upskilling by 2030; that investment lands only when there is a redesigned role to receive it.
5. Scale the tool last. Once the work has somewhere to land, widen the deployment. This is the order that turns the people-and-process work into the lever it is meant to be.

The US Census Bureau's Business Trends and Outlook Survey, the government's own primary count of how firms use AI, shows where most organisations still sit: by mid-2026 only 17% to 20% of US businesses reported using AI, rising to roughly 37% among firms of 250 employees or more, and the same data shows that the firms integrating AI more deeply across functions report measurably stronger commercial performance. Breadth of integration tracks with results; surface-level adoption does not. The leaders who pull ahead are not running better models. They are running redesigned work. Phase One was muscle. Phase Two was machine. Phase Three is mind: the win goes to the leaders who reshape the organisation around the technology, then upgrade the way their people think and decide, rather than expecting a tool to do that work for them.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.