New Work Is Dead? Great. Now We Can Do the Real Thing.

For a while, “New Work” sounded like an unstoppable movement.

Then the pendulum swung:

• return-to-office mandates

• cost pressure

• fewer perks

• more “efficiency”

• and, of course, AI everywhere

So yes—New Work feels like it’s shrinking.

But I think the more honest question is this:

Was New Work ever what we told ourselves it was?

Because if your New Work program can be canceled by removing the free snacks and the “remote-first” slogan, then—awkward—it wasn’t New Work. It was interior design with HR messaging.

The good news: the real version of New Work is not dead. The bad news: it’s also not “comfortable”.

And AI is about to make that painfully obvious.

The uncomfortable truth: we turned New Work into lifestyle branding

In too many organizations, New Work got reduced to:

• hybrid policies as identity politics

• benefits as culture strategy

• “trust” as a poster, not a management system

• self-organization… without decision rights

New Work didn’t fail because flexibility is wrong. It failed because we treated surface-level signals as the transformation.

Ironically, research shows hybrid work can work very well when it’s actually managed: a randomized control trial at Trip.com found people working from home two days a week were just as productive, just as likely to be promoted, and resignations dropped by 33% for employees moving from full-time office to hybrid. (SIEPR)

So the problem isn’t “remote vs office”. The problem is: how the organization makes decisions, distributes responsibility, and learns.

The original idea behind New Work was never the table football

The term “New Work” is credited to Frithjof Bergmann—and at its heart was a pretty radical thought:

Use technology to reduce boring, repetitive work so humans can do more creative, meaningful work. (context.org)

That idea aged well.

Because now we have AI.

And AI is doing exactly what Bergmann was pointing at:

• producing drafts

• suggesting options

• writing and summarizing

• generating code, content, and analysis

The question is: what do we do with the freedom it creates?

That’s where most companies get stuck.

AI doesn’t just change tools. It changes the economics of work.

Two data points that should make leaders slightly nervous:

• 75% of knowledge workers already use AI at work (global survey, Microsoft/LinkedIn Work Trend Index 2024). (Azure CDN)

• 78% of AI users are bringing their own AI tools to work (BYOAI)—often without guidance or clearance. (Azure CDN)

Translation: AI is already in your organization—whether procurement approved it or not.

Now here’s the part many underestimate:

When drafting becomes cheap, judgement becomes expensive.

The scarce capability shifts from “creating the first version” to:

• framing the right problem

• deciding what matters

• validating correctness and risk

• integrating safely into a living system

• owning consequences

In short: decision-making becomes the bottleneck.

And if your operating model still relies on centralized approvals and hierarchical escalations… congratulations: you just built a world-class bottleneck generator.

AI will feed it beautifully. Every day. At scale.

“But AI will make us faster.” Not automatically.

There’s a strong temptation to treat AI like a productivity cheat code.

Reality is more nuanced.

Yes, AI can improve productivity. A well-known field study in customer support found that introducing a generative AI assistant increased productivity by about 14–15% on average, with much larger gains (up to ~34%) for less experienced workers. (NBER)

But there’s also a warning sign from software delivery research:

The 2024 DORA research highlights that higher AI adoption, in their model, was associated with a ~1.5% decrease in delivery throughput and a ~7.2% reduction in delivery stability. (Google Cloud)

Read that again: more AI, slightly less throughput, and noticeably less stability.

That doesn’t mean “AI is bad.” It means: AI amplifies the system it’s plugged into.

If your fundamentals are weak (quality discipline, testing, small batch sizes, clear priorities), AI will help you produce more… instability.

So what does “New Work” mean in the AI era?

It becomes less about perks and more about organizational design.

Here are three shifts that matter (and none of them fit on a culture poster):

1) Decision rights beat “empowerment” slogans

AI creates options fast. The limiting factor becomes who can decide and how quickly.

Teams need real decision latitude:

• what to build next

• how to solve it

• what “good” looks like

• when something is safe enough to ship

Not “we trust you” followed by a 12-step approval chain.

A pragmatic test: How long does it take in your org to make a meaningful product or engineering decision? If the answer is “it depends who’s in the meeting,” you don’t have empowerment—you have astrology.

2) Validation discipline becomes a first-class capability

AI output is not a deliverable. It’s a proposal.

In AI-enabled work, the team’s craft shifts toward:

• stronger peer review

• intentional testing

• traceability (“what changed and why”)

• security-by-default

• clear “definition of done” for AI-assisted artifacts

If you skip this, you will ship confidently wrong things faster—which is… a strategy, I guess.

(And yes, AI-specific risks are now mainstream enough that OWASP maintains a “Top 10 for LLM Applications”. Neglecting validation is literally on the list of things that go wrong. (OWASP Foundation))

3) System stewardship becomes non-negotiable

AI increases change volume. That increases operational risk.

Organizations that thrive invest in:

• internal platforms that make the right thing the easy thing

• observability and fast feedback loops

• guardrails (not bureaucracy)

• release quality and reliability engineering

This is where “New Work” gets very unromantic, very fast. But it’s also where competitive advantage shows up.

Leadership doesn’t disappear. It gets harder.

AI doesn’t eliminate leadership. It removes excuses.

In high-performing environments, leadership becomes:

• direction (what outcomes matter)

• context (constraints, trade-offs, priorities)

• enablement (capability building, platform investment, removing friction)

• accountability (owning system-level consequences)

Not micromanagement. Not abdication. Clarity.

Because “self-organization” without clarity is just… people being busy in parallel.

The real paradox: AI adoption with old leadership models

Many companies are introducing AI while keeping leadership models from the “command-and-control” era.

What happens then is predictable:

• AI increases output

• the organization increases controls

• teams lose autonomy

• decision latency increases

• people get frustrated and start using unsanctioned tools anyway

• quality and trust erode

That’s not transformation. That’s churn with better autocomplete.

And the Work Trend Index data suggests exactly this dynamic: employees aren’t waiting for official rollout—they’re already using AI, often bringing their own tools. (Azure CDN)

If you don’t provide safe, supported ways of working with AI, you’ll get shadow usage by default. Not because people are malicious—because they’re trying to get work done.

A pragmatic checklist for leaders (no buzzwords, sorry)

If you’re serious about “New Work” in an AI world, answer these honestly:

1. Where are decisions made—actually? Do teams have real decision rights, or just responsibility without authority?

2. What is your standard for validation? What must be true before AI-assisted work ships (tests, reviews, security checks)?

3. What are your guardrails? Are they enabling speed—or just rebranding control?

4. Do you provide a sanctioned AI path? Tools, training, and policy that help people move fast without leaking data.

5. What do you measure? Output metrics will reward noise. Outcome metrics reward learning and value.

My conclusion

New Work isn’t dead. The watered-down version is.

The AI era makes the original idea unavoidable:

• Decisions must move closer to where knowledge sits

• Validation must be stronger, not weaker

• Leadership must provide clarity, not control theatre

• Learning speed becomes the real differentiator

Or in one line:

The future won’t belong to the most efficient organizations.It will belong to the ones that learn fastest—without breaking production.

Question for you

In your organization, is AI currently:

• accelerating learning and decision-making, or

• accelerating the volume of things that need approval?

And what do you believe wins in the AI era: more control—or better decision capability with clear guardrails?