Restructuring for the Agentic Age

The old assumption has broken

A few years ago, most leaders still saw organisational restructuring as a rare, painful event: you grit your teeth through one big reorg, ride out the disruption, and then coast in a stable structure for years.

That comfortable assumption is now obsolete. AI, especially agentic tools that do not just assist but actively execute, has turned restructuring into a permanent, ongoing discipline. Roles, team ratios, skill mixes, and even the very definition of “output” now shift on a quarterly cadence. Tech leaders who treat reorgs as occasional housekeeping are setting themselves up for chronic lag behind competitors who have already accepted the new reality: the org chart is no longer a static blueprint; it is a living system that must evolve in real time.

The first quarter of 2026 made it obvious

At Block, CEO Jack Dorsey was unusually direct. After cutting more than 4,000 roles in late February 2026 and taking the company from more than 10,000 employees to just under 6,000, he argued that a significantly smaller team can now do more and do it better with intelligence tools. More telling than the numbers was the rationale: better to make one decisive move than pretend AI has not changed what efficient organisational size looks like.

At Atlassian, CEO Mike Cannon-Brookes was similarly explicit in a March 11 team update. He tied a ~10% headcount reduction, roughly 1,600 roles, to self-funding AI acceleration and enterprise sales, and wrote that it would be “disingenuous” to pretend AI does not change the mix of skills required or the number of roles needed in certain areas.

At Oracle, co-CEOs Clay Magouyrk and Mike Sicilia represent the other side of the same equation: stepped-up AI infrastructure investment alongside continued cost pressure and reports of layoffs tied to that buildout. At Workday, then-CEO Carl Eschenbach’s February 4 8-K disclosed a reduction of approximately 2% of the workforce, primarily in non-revenue-generating roles, while the company continued to hire in strategic and revenue-generating areas.

At Pendo, CEO Todd Olson called the company’s ~10% cut, 90 roles, part of “refounding” the business. His explanation matters: customers are now building agents, building with agents, and changing how they work with AI. That is not defensive language. It is a direct admission that customer behaviour is already forcing operating-model change.

Engineering feels the shift first

The engineering organisation is feeling the shift first and hardest.

At Meta, CEO Mark Zuckerberg is pushing similar logic into internal engineering. Reuters reported that top software engineers were drafted into a new Applied AI Engineering unit, and that joining was “no longer voluntary.”

On the REWORK podcast, 37signals co-founder and CTO David Heinemeier Hansson described the move from AI as editor-autocomplete to full agent mode: agents in the terminal, wielding tools, executing commands, and iterating in tight feedback loops. The point is not the spectacle. The point is that the ceiling on individual output is moving fast.

Snowflake CEO Sridhar Ramaswamy has been pushing a similar efficiency logic while the company makes targeted adjustments to staffing. And at C3.ai, CEO Stephen Ehikian’s February restructuring cut roughly 26% of the global workforce and targeted about 30% lower annualised non-employee costs. Efficiency is no longer a one-time project. It is table stakes.

What leadership is really trying to answer

Recent restructurings show these questions have left the whiteboard and entered the real world.

The dangerous governance gap

Gartner’s January 2026 future-of-work guidance for CHROs labels one of the defining risks of the moment “Reductions in Force (RIFs) Before Reality.” Some CEOs are cutting headcount on optimistic assumptions about AI returns that have not fully materialised yet. Others are simply reacting without the data needed to do it well. Either way, CHROs are being asked to lead “talent remix” efforts so workforce size and structure actually match strategy in a sustainable way.

My stronger take: there is real opportunity here, but also real risk of overshooting. It is dangerously easy to build a spreadsheet-perfect org that looks lean and disciplined but collapses under real pressure because coverage, context, and team cohesion were treated as disposable line items. AI productivity is not magic. It still needs humans who understand the bigger picture.

The real problem: fragmented data and static thinking

The cadence of change has accelerated dramatically, yet most organisations still make restructuring decisions with siloed, outdated information:

• HR systems own reporting lines and titles
• Engineering tools approximate output
• Finance systems track cost
• AI platforms hold usage, adoption, and the earliest signals of genuine productivity gains

Because these views rarely connect, leaders are forced to bet blind. Restructuring becomes a series of expensive real-world experiments instead of low-risk scenario testing.

The smarter path: model your organisation as a connected system

The fix is straightforward in concept, powerful in practice. When you treat the organisation as a single, connected model, linking teams, roles, costs, coverage, delivery constraints, and AI usage signals, you can explore structural options before anyone’s role changes.

You can ask and answer:

• How does a manager’s span of control scale when individual contributors become 3-5x more productive in certain workflows?
• Does shifting from function-based to outcome-based teams unlock throughput under heavy agent assistance?
• Which roles turn into new bottlenecks when agents crush execution volume but spike demand for review, validation, and integration?

This is the new leadership imperative. AI does not just force faster change; it gives us the tools to simulate that change intelligently. Leaders who embrace proactive modelling will make sharper, less traumatic decisions. Those who do not will keep lurching from one reactive reorg to the next.

From episodic reorgs to continuous design

Judgement, culture, and human dynamics will never be fully modelled. But they no longer have to be the only inputs. Restructuring stops being an episodic crisis and becomes a continuous, evidence-driven capability, with shorter learning cycles and far higher confidence.

Tools like Orgonaut are being built exactly for this moment, giving leadership teams a practical way to explore structures, surface trade-offs, and move forward with clarity when they finally decide to act.