Most of us won’t lose our jobs to a single AI tool. But we may find ourselves slowly pushed out of the workflows we’ve always been part of. The shift will come quietly, through pilot projects that introduce AI agents, not to replace us outright, but to unbundle our roles and reassign tasks in ways that reshape what our jobs even are.

We’ve entered a new phase of AI transformation. It’s no longer only about adopting tools; it’s about completely rethinking how labor flows through organizations: the Great Rewiring of knowledge work.

Many companies are still officially “experimenting” with AI. But they are already pre-emptively freezing hiring and cutting roles in anticipation of efficiencies AI might unlock in the near future. We see the pattern across sectors. Banks like ABN AMRO are planning to shed thousands of roles in the name of simplification and end-to-end digitization. Retail and logistics players such as Amazon, Target, and UPS are cutting corporate and management layers while investing heavily in automation and “new operating models.” The big consulting firms are shrinking the junior pyramid and aiming for a diamond-shaped workforce, hiring fewer graduates while searching hard for AI and data talent. Tech giants like Microsoft are trimming management layers and “reallocating resources” even as they double down on Copilot and other AI-first products.

These moves are not the result of proven automation outcomes—they’re strategic bets on a future that hasn’t fully arrived yet.

This is our 2026 (and beyond) outlook on the future of work — and a hopeful call to action for those ready to lead. As we head into the holidays, it’s worth remembering: the future of work isn’t written yet. We get to help shape it, thoughtfully and humanely.

We keep telling a very comforting story about AI and work

The problem is that it quietly assumes something that is no longer true: that the world of work stays more or less the same.

Step 1: Individual productivity with AI tools

Nothing we can argue with, if you ignore one basic fact. We are not solo performers. We sit inside a chain of people, systems, and approvals. We hand work to others, they hand work back to you, and the whole mess eventually produces value for a customer.

If I become three times faster but the chain around me stays the same, the effect on the system is small. Extra speed and innovation quietly dissolve into the sludge of meetings and legacy tools. This is why companies today see little ROI on their AI tool investments, despite improving adoption.

Institutions like the OECD and Goldman Sachs now assume that a quarter or more of current tasks could in principle be automated, and that most jobs will be transformed rather than “augmented”. In other words, the numbers are already pushing companies toward workflow redesign, not just individual productivity hacks.

Step 2: Atomizing workflows with AI agents

Think of the work in your organisation as a molecule. Inside that molecule are smaller atoms—the individual tasks. For decades, we have glued certain atoms together and called that a job. Accountant. Recruiter. Lawyer, Sales Director. Product Manager.

AI does not just make each job a bit faster. It enables breaking the entire thing apart and rebuilding it as a kind of knowledge-work factory where agents work alongside human operators.

Instead of merely “rolling out Copilot,” serious companies choose a high-impact workflow, and then do something far more radical:

1. Map how things really flow today, end-to-end, fearlessly crossing functional silos

2. Break the work into small tasks such as classify, extract, check, route, decide, and explain

3. Ask which of those tasks can be handled by agents, which by rules, and which still need humans

4. Redesign the workflow around AI agents as the default, and humans at the critical junctions

The result isn’t just faster work. It’s different work.

The new shape of work

As organizations begin to atomize and redesign workflows, they will need three distinct types of people: architects, who help design the new workflows; a larger group of operators, who run the system and manage human relationships; and everyone else, whose tasks are reduced to automated atoms.

“Operator” sounds mechanical, like someone simply pushing buttons. But for now, it’s a placeholder for something deeply human: the people who keep systems working not just technically, but socially. Operators bring oversight, accountability, and the judgment to step in when automation falls short, AND also manage relationships, build trust, and handle the nuance that machines still can’t touch.

Managing talent

Entry-level roles aren’t just grunt work that AI can replace. They’re the places where future architects and operators build domain expertise, judgment, and systemic awareness. If we cut too fast, we gut the very talent we’ll soon realize we need.

The solution isn’t to keep everyone. It’s to manage talent deliberately for the future of agentic AI:

• Identify who shows architect or operator potential

• Develop them intentionally for the redesigned system

• Reduce roles that are truly just automated atoms

• But protect the pipeline of future leadership

Organizations already seeing bottom-line gains from AI aren’t just adopting tools. They’re cultivating the human capabilities required to lead the transition, before the workflows rewire around them.

Leading the Talent, Not Just the Tech

Senior leaders must personally take up the AI agenda: model its use, fund it with real budgets, and set clear boundaries on where humans must stay in the loop. They need to move beyond tool adoption and start shaping how AI integrates into everyday processes, not just to cut costs but to strengthen trust, resilience, and long-term capability.

Leading means protecting the human layer, not squeezing it in the name of short-term efficiency. The operators and architects of tomorrow don’t emerge from nowhere. If you automate away the wrong roles without building new ones, you’re not leading a transformation; you’re hollowing out your future.

Owning the system also means co-owning it across functions. No leader can work in isolation. The new workflows will span IT, data, risk, HR, and finance — and leadership teams will need to make joint calls on where agents fit, what risks are tolerable, and how to reassign people whose roles are being phased out. “My function first” is not a serious stance in this world.

When will the agentic AI rewiring happen?

It’s also political, shaped by siloed power structures and informal relationships, with no two organizations running workflows in the same way.

Early waves will hit:

• High volume, lower stakes workflows (customer service, basic analysis, document processing)

• Companies with clean data and well-documented processes

• Functions where output is easily measured (sales ops, marketing ops, financial reporting, coding)

Later waves will hit:

• Complex, high-stakes workflows (strategic planning, major negotiations, research and development)

• Companies with messy legacy systems (if they still exist)

• Functions where judgment and relationships dominate

What this really means for organisations in 2026

That means identifying which workflows matter most, spotting who’s already adapting, automating, and reshaping things quietly, and putting those people in real positions of responsibility.

We’ve seen what happens when organizations cut too fast, train too generically, or leave workflow redesign to someone else’s function.

No one has the complete blueprint, but the longer you wait, the harder it gets. Early movers are learning what works and what breaks. Companies like Walmart and BNY aren’t just playing with agents anymore. They’re already seeing tangible results: shortened product timelines, increased capacity, where digital agents function as team members, complete with their own access, oversight, and roles.

Every week, they’re gaining capabilities that will be hard to catch up with later.

We help companies identify the climb, not with detailed roadmaps, but with a clearer view of the terrain ahead. We work with you to spot real talent, prioritise workflows, and learn from actual progress, not polished slides.

Our upcoming Christmas podcast dives deeper into how companies can lead this shift. And in the new year, we’ll break down the ideas here into focused, practical posts to guide transformation in 2026 and beyond.

Thanks for reading futurebraining! Subscribe for free to receive new posts and support our work.