Issue 007: What Leaders Think They See

IBM told the market it would freeze hiring for 7,800 jobs because AI could do them. By 2024, the company’s overall headcount had grown again anyway. Salesforce’s CEO promised the public that AI would mean "radical augmentation," not job losses, then went on a podcast a month later and said "I need less heads." Duolingo’s CEO wrote a memo about going AI-first and had to walk half of it back within weeks once his own employees read it. Something keeps going wrong in the gap between what leadership thinks an AI rollout looks like and what the people living through it are actually experiencing.

This week is about that gap. Five real companies getting it badly wrong, and one getting it right, as they roll AI into their workforce. Three frameworks for the specific ways that gap opens up between a leader’s dashboard and a team’s reality.

Stephanie Hills is this week’s Leader Spotlight. She spent 25 years inside three Fortune 500 tech leadership teams and now advises executives on exactly this problem. She joins The Work Life Reporter Live this Tuesday to talk about what to do when the role you’re leading through keeps shifting under you.

Welcome to the seventh issue of The Work Life Reporter. This week you get:

Let’s get into it.

In May 2023, IBM CEO Arvind Krishna announced the company would pause hiring for roughly 7,800 back-office roles he expected AI could handle within five years, mostly HR functions like payroll queries, leave approvals, and documentation. IBM built an AI assistant, AskHR, that did exactly what it was built for: it handled 94% of those standard HR tasks cleanly.

The read from the dashboard said the rollout was done. The read from the floor was different. The remaining 6% of queries, the ones needing human judgment or empathy (a leave request tangled up with a family emergency, a payroll error a chatbot couldn't parse) started stacking up as real operational gaps. By 2024, IBM's overall workforce had grown again, not shrunk, as the company moved the savings from automation into roles AI couldn't touch: software development, sales, client relations. What looked like a finished rollout on a dashboard was, on the ground, an unfinished one that still needed people to close it out.

In August 2025, Salesforce CEO Marc Benioff told an AI summit that AI wouldn't cause a white collar wipeout, promising “radical augmentation” instead of replacement. One month later, on a podcast, he confirmed Salesforce had cut its support division from 9,000 people to about 5,000 because Agentforce, the company's AI agent product, now handles half of all support interactions. His words: “I need less heads.”

Both statements might even be true at a portfolio level. But said back to back, in public, they didn't read as two true things, they read as one leader saying whatever suited the room he was in. The employees who left weren't reading an internal memo with more nuance attached, they were reading the same quote everyone else did. The mechanics of the AI rollout itself appear to have worked. What failed was that the public language used to describe the same change contradicted itself within weeks, and the contradiction became the story instead of the technology.

In April 2025, Duolingo CEO Luis von Ahn sent an “AI-first” memo announcing the company would track employee AI use in performance reviews, phase out contractors AI could replace, and only approve new headcount once a team proved it couldn't automate the work. The backlash was immediate and public. “AI first means people last” became one of the more shared reactions.

Von Ahn told the Financial Times he “did not expect the amount of blowback.” Within weeks he reversed the performance review policy and told a podcast, “when I released my AI memo a few weeks ago, I didn't do that well,” clarifying Duolingo was still hiring at its usual pace. What broke wasn't the ambition, plenty of leaders want AI adoption to be real and measured. What broke was sequencing: he announced a policy that would be felt by every employee before testing how it would land, then had to rebuild trust in public, which costs far more than getting the first version right would have.

In July 2024, Intuit CEO Sasan Goodarzi told employees the company was cutting 1,800 roles “to increase our investment” in AI, while simultaneously hiring roughly 1,800 new people into engineering, product, and customer-facing roles built around it. In May 2026, when Intuit cut a further 3,000-plus roles, about 17% of headcount, Goodarzi told CNBC the cut had “nothing to do with AI.”

Both statements are individually defensible. Workforce strategy is genuinely complicated and layoffs rarely have one clean cause. But an employee who remembers the 2024 memo, and remembers it framed AI investment as the explicit reason for that round, hears the 2026 denial as a walk-back, not a clarification. Once a leader has explicitly linked layoffs to AI investment once, denying the link the next time doesn't read as more honest. It reads as managing the headline.

In January 2024, SAP announced a restructuring touching about 8,000 roles, roughly 7% of its workforce, as it shifted focus toward AI. Instead of leading with cuts, SAP built the programme mostly around voluntary leave and internal reskilling, backed by a stated €2 billion investment, and said publicly it expected to exit the year at a headcount similar to where it started.

The distinguishing choice wasn't the restructuring itself, plenty of companies were making similar calls in 2024. It was where SAP put its money and its language: reskilling and voluntary programmes ahead of straight cuts, and a headcount target specific enough to be checked against later. That's a harder commitment to make than a vague “we care about our people” line, because it's falsifiable. Leaders willing to be checked tend to be believed.

Prudential's 2026 Future of Work study found 44% of employers see AI as “extremely or very impactful” to their roles, against just 21% of employees. Positivity splits even further: 78% of employers view AI positively, against 51% of employees, a 27-point gap. WalkMe's 2026 State of Digital Adoption report found only 9% of workers trust AI for complex, business-critical decisions, against 61% of executives, and that 54% of employees quietly bypassed their company's AI tools in the past 30 days and did the task manually instead.

That's not a rollout with a few stragglers. That's two different experiences of the same tool, both being reported up the same chain as one experience.

The dashboard a leader sees, usage numbers, adoption rates, sentiment scores rolled up into one figure, is almost never the same thing a team is living. Stop treating adoption metrics as a proxy for trust. Ask directly, and separately, whether people believe the tool works for the hard cases, not just whether they clicked on it this month.

Gallup's 2025 State of the Global Workplace found less than half the world's managers, 44%, have received any formal management training. The same research found managers who did receive training in effective coaching saw their own performance metrics improve 20 to 28%, and the teams they led saw engagement rise by up to 18%.

Most organisations train people to manage the mechanics of change: the announcement, the timeline, the FAQ document. Almost none train them to lead a team through the part that isn't on any timeline, the weeks where trust is either being rebuilt or quietly spent down, one interaction at a time.

If your leadership development budget goes toward change management process and nothing toward how managers actually talk to people mid-change, you're training half the job. The 44% figure isn't a training gap. It's a trust gap wearing a training gap's clothes.

Businessolver's 2026 State of Workplace Empathy report found only 68% of HR professionals now rate their CEOs as empathetic, down 16 points from 2022 and the lowest score since the survey began in 2016. Meanwhile, Challenger's tracking shows AI was cited as the reason for 101,743 job cuts in the first half of 2026 alone, nearly double the 54,836 cited for all of 2025. Among employees who are worried about their own role being reduced or eliminated by AI, 49% say extra effort feels pointless, against 27% of employees who aren't worried.

That's the actual cost of a badly handled AI announcement. It isn't measured in headlines. It's measured in the discretionary effort that quietly stops showing up from everyone still on the team, not just the people who left.

Every AI-linked restructuring announcement is also, whether intended or not, an empathy test broadcast to everyone who's still there. Plan the announcement with that in mind, not as an afterthought once the numbers are settled.

Stephanie Hills spent more than two decades in senior engineering and technology leadership roles across three Fortune 500 companies, Cisco, NCR, and Scientific Atlanta. She now works as an executive advisor focused on engineering change and AI readiness through Hills Coaching and Consulting.

Since then, she’s sat inside three separate Fortune 500 engineering overhauls and watched, from the inside, what actually decided whether a team held together or quietly checked out, built her advisory practice specifically around the gap between what a leadership dashboard shows and what a team is actually living through during AI-driven change, and made owning the role you’re in as central to her framework as making the bold move, for leaders whose job keeps shifting under them faster than anyone can retrain for it.

In this week’s Leader Spotlight, we sit down with Stephanie to unpack the gap between what leadership dashboards show and what teams are actually living through, the mistake tech leaders keep making when they roll out AI, and what owning the role you’re in means when the role itself won’t stop shifting. A note on what follows: these are drawn from the discussion points Stephanie has confirmed for Tuesday’s conversation and her own public positioning, not a transcript, since the conversation itself hasn’t happened yet as this issue goes out. Come back after Tuesday for her actual words.

Adoption rates and rollout metrics tell a leader a story about completion. Stephanie's angle for Tuesday is what that story leaves out, and why the leader watching the dashboard is often the last person in the building to notice what their team is actually experiencing.

Drawn from three Fortune 500 engineering overhauls, Stephanie's read on timing and communication: the technical rollout and the trust rollout run on different clocks, and treating them as the same thing is where the cost shows up later.

Most leadership development covers the mechanics: the announcement, the plan, the timeline. Tuesday's conversation digs into the part that isn't on any timeline, the ongoing work of leading people through change you don't fully control.

For leaders whose job description is being rewritten in real time by the technology they're deploying, Stephanie's advice isn't about waiting for the ground to stop moving. It's about what to do while it still is.

Not a framework nobody uses. Tuesday's conversation closes on something a people leader can act on this week, not next quarter.

Thank you in advance, Stephanie, we’ll see you Tuesday!

Connect with Stephanie on LinkedIn or find out more about her work at Hills Coaching and Consulting

And that’s a wrap for Issue 007. Something land? Hit reply and tell us which section you’re taking back to your team.

Follow The Work Life Reporter on LinkedIn for what we share between issues, and join us live on Tuesday. Visit theworklifereporter.com for more.

PS. Know someone who leads people and would find this useful? Forward it on. Ten seconds to share, a lot longer to write.

PPS. Got a story, a leader, or a topic you'd like us to dig into? Reply with your ideas.