The Turn: The AI Talent Battle That Isn't

Throwing cold water on the latest "talent war"

We’ve developed a bad habit of treating individual talent moves like market research.

Someone notable leaves one software company for another, and within hours, there’s a hot take about what it means for the incumbent’s competitive position, its AI credibility, and its long-term viability.

The more senior the person, the more definitive the conclusion for some reason.

It’s gotten to the point where every hire announcement from Anthropic or OpenAI comes pre-loaded with a narrative about who’s winning and whose goose is getting cooked.

This is not a useful way to read the industry.

People move for a lot of reasons. It can be a role that turned out to be the wrong fit, a challenge they wanted to explore, or a working environment that suited how they actually want to work. There can be personal reasons or preferences, even for big-time executives.

None of those reasons are completely invisible in a job announcement, but none of them are completely visible either. What you see is a name, a title, and a press release written by a comms team.

I wrote on Reworked this week about Peter Bailis leaving Workday for Anthropic, partly because the coverage got some things meaningfully wrong, and partly because it’s an example of this broader reflex. From misreading what the title actually means to stretching the competitive angle, too much got skipped.

None of that is surprising. Nuance doesn’t travel well in a news cycle. But the cumulative effect of treating every notable hire as a verdict is that we end up with a lot of confident conclusions built on thin evidence.

The talent patterns that actually tell you something are aggregate, not individual.

• Which kinds of people are choosing to build at the model layer versus the application layer?

• Is the flow directional and sustained or scattered?

• What do dozens of moves reveal about where the hard problems are perceived to be?

One person’s career decision, for any number of reasons you’ll never fully know, doesn’t get you there.

We’re going to keep seeing these hires, and they’re going to keep getting this treatment. Don’t read too much into it on an individual level. Look for larger trends. I know it’s less fun, though.

Use AI to research in a way that’s smart

I’ve been annoyed by bad research since before Wikipedia and “Source: Google” citations.

The most annoying part is that people today have much more powerful tools at their disposal, yet they use them to synthesize the most boring conclusions. “AI is disrupting recruiting,” or “Entry-level jobs are changing.”

Wow. I definitely needed a summary on that.

I recently shared how I structure my general research process using AI. It is NOT a pretty process, and it is expensive as far as usage is concerned. But instead of a Reader’s Digest version of the news, it gives you an organized reading list to get you an in-depth read on what has and is happening.

What else is happening this week?

• Meta AI Lets Zuckerberg’s Staff Talk to the Boss — Meta built an AI clone of Zuckerberg so all 79,000 employees can feel connected to a CEO who apparently couldn’t manage it himself. Nothing says “we value our people” like a chatbot wearing the boss’s face.

• Gallup: Rising AI Adoption Spurs Workforce Changes — Half of workers use AI at least sometimes, but only one in ten says it’s actually changed how work gets done. Companies that adopted it are somehow hiring more and cutting more simultaneously, in case you were wondering who’s pocketing the productivity gains.

• Is Joy at Work Possible? — Laurie Ruettimann talked to Yolanda Fraction, who argues joy is more achievable than happiness at work because it doesn't require things to be going well. A genuinely useful distinction most workplace wellness gurus never bother to make.

• Zoomers Are Sabotaging AI at Work — Nearly half of Gen Z workers are deliberately breaking their company’s AI rollouts by leaking data, using unapproved tools, and submitting garbage output. The researchers recommend better change management rather than stopping the automation victory laps while threatening people’s livelihoods.

• Minimum Wage Hikes Are Quietly Raising Injury Rates — When the minimum wage goes up 10%, companies respond by working people harder rather than trimming headcount, producing two to three additional injuries per 1,000 workers annually. You got a raise and a repetitive stress injury.

• Workplace Management Tech Is Everywhere, But Integration Challenges Limit Its Value — Workplace management tech is widely deployed and barely anyone can get it to talk to their other systems. This sentence has been true of enterprise software since 1998 and probably won’t change with AI.

• Bosses Are Getting Worse. AI Is Part of Why. — Managers are outsourcing their judgment to ChatGPT, and their employees are cleaning up after it. Groundbreaking findings from the year 2026.

• The New Architecture of Work — AI is collapsing the cost of expertise to near-zero and most enterprise software was built for organizational structures that no longer make sense. Jason Corsello proposes replacing static job titles with dynamic task categories, which is the right idea and one nobody will actually implement.

• Should AI Agents Be Treated Like Colleagues? — A journalist staffed a startup entirely with AI agents and was stunned when they argued, disobeyed, and questioned his authority. The answer to the headline is no, and 82% of deploying companies are going to learn that firsthand.

• The Remote Work Fantasy Breaks in Real Life — A designer who moved through nine countries concludes that stability produces better work than novelty. The files always come with you.

• Could Microsoft Win the War for Enterprise AI? — While everyone obsesses over OpenAI and Anthropic, Josh Bersin argues Microsoft is quietly positioned to capture the enterprise AI market through Office, Azure, and existing customer relationships. The companies that control distribution usually win. Or at least that’s what the guy who poured me my IPA last night said.

• The Future of Work Is Here, But Hiring Hasn’t Caught Up — AI and automation have reshaped what companies actually need from workers, but recruiting is still running the 2019 playbook. The gap between how work has changed and how companies hire for it is where a lot of current talent pain lives.

• Zoom CEO Says 3-Day Workweek Could Be the Future — Eric Yuan, whose company nearly collapsed when offices reopened, is very excited about AI making the five-day week obsolete. A real study backs him up on the time savings, but he’s a CEO, so the question of who actually keeps those hours went unaddressed.

• The Agent Illusion — Most things vendors are selling as AI agents are just workflows with a better interface. Jason Averbook’s test for a real agent: Does it adapt to new conditions or just follow a predetermined path, and does it remember anything? This also might work on me.

• The Missing Layer in AI Transformation — CHROs are deploying AI without first getting clear on what work they’re trying to change. This is also what happened with every ERP, HCM, and workforce planning platform before this one.

Have a great rest of your week!