I’ve spent the last few weeks watching three of the biggest tech companies in the world launch what they’re calling AI “coworkers” within roughly a five-month window, and I keep coming back to the same uncomfortable thought.
Nobody launching these products is being asked the question that actually matters to the person reading about them on their phone during a commute, wondering if their job still exists in six months.
Let me start with what actually shipped. OpenAI launched ChatGPT Work on July 9, powered by its new GPT-5.6 model, folding its Codex coding tool and a persistent cloud agent into one desktop app while quietly sunsetting its standalone Atlas browser.
Anthropic got there first with Claude Cowork, which went general availability back in April and now ships inside every paid Claude plan, running as a third tab beside Chat and Code, reading your files and executing scheduled tasks while you’re in a meeting.
Microsoft’s Copilot Cowork reached general availability in mid-June, plugged directly into Word, Excel, Outlook and Teams, and it’s already inside more than half the Fortune 500.
The Benefits Are Real, Not Marketing Spin
These companies aren’t reporting hypothetical wins. Microsoft says one engineering team taught Cowork to compare nearly four thousand files across two product versions, work that would have taken weeks, done in a fraction of the time. A sales team pointed it at a stalled pipeline and got back a ranked list of dead deals with the exact follow-up message needed to revive each one, collapsing a week of manual review into a morning. Salesforce says its AI agents now handle roughly half of all customer support interactions. These are measured outcomes, not vendor optimism, and that’s exactly why the layoffs follow so quickly behind them.
Salesforce cut 4,000 support roles after deploying that same agent technology, then cut 1,000 more this year. IBM has replaced close to 200 HR positions with agents while tripling its entry-level AI hiring. Amazon has cut roughly 30,000 corporate roles since late 2025. Tech layoffs are running at over 1,100 a day in 2026, nearly double last year’s pace and companies are increasingly comfortable naming AI as the reason out loud, something that would have been a PR disaster eighteen months ago.
The Problem Nobody Is Willing to Say Plainly
The actual data on whether AI caused these layoffs is a mess, and that ambiguity is being used against workers. Challenger, Gray & Christmas attributes somewhere between 8% and 26% of layoffs to AI in any given month. Nikkei Asia puts the tech-sector share closer to half. Sam Altman himself has admitted there’s “AI washing,” where companies blame AI for cuts they’d have made anyway. Oxford Economics found firms aren’t replacing workers with AI at anywhere near the scale their announcements imply. Marc Andreessen calls AI “the silver bullet excuse” for correcting pandemic-era overhiring.
Nobody agrees on the real number, and that’s precisely the point, when the reason is unfalsifiable, no worker can tell if their specific role is genuinely obsolete or just conveniently reclassified.
The $1 Billion Question Nobody Is Asking
HP told investors its AI-driven headcount cuts would net $1 billion in savings by 2028. That’s one company, saying the number out loud. Multiply that logic across Citigroup, SAP, Dow, PayPal, and every other firm running the identical playbook, and you get tens of billions in reallocated payroll, money that isn’t going to retraining, isn’t going to the workers who lost the roles, and in most cases isn’t even landing on the bottom line as profit. It’s going straight into GPUs. Meta, Amazon, Microsoft and Alphabet alone are spending roughly $700 billion this year on AI infrastructure.
The billion-dollar question isn’t “will AI take my job.” It’s: when it does, where does that saved money actually go, and why does none of it flow back to the people whose labor funded the productivity gain in the first place?
Enter the Agentic Fleet Manager
Here’s the opening most people are missing entirely. Every company deploying ChatGPT Work, Claude Cowork, or Copilot Cowork at scale needs someone who isn’t just using the agent, but running a fleet of them, assigning tasks, checking output quality, managing which agent touches which system, and owning the outcome when something goes wrong. That role barely exists yet as an official title, which means the person who builds those skills first isn’t competing for a shrinking job. They’re stepping into one nobody’s filled.
The Solution: How to Actually Get There
Map your fear to tasks, not titles. These tools automate repeatable, single-tool actions like drafting, formatting, and scheduling, not judgment calls, relationships, or accountability. Know which of your tasks fall into which bucket, because the accountability tasks are the ones that turn into fleet-manager responsibility.
Get inside one of these tools before you’re forced to. IBM’s entry-level AI hiring tripled even as it cut other roles, meaning the demand moved, it didn’t vanish and the people running multiple agents day to day are the ones being pulled into that demand first.
Become the verifier. Every one of these systems still needs a human who can catch when a finished deck or spreadsheet is confidently wrong. That verification habit is the exact skill that scales into managing several agents at once instead of just one.
Attach yourself to cross-functional work. Narrow, single-tool, repeatable jobs are exactly what’s being cut first, spread across tools and context instead, which is precisely the vantage point a fleet manager needs.
Quantify your own output. Track your results in numbers your company already measures, so when the conversation turns to “who oversees our agent deployment,” you’re the person with a documented track record, not a guess.
Do these five things consistently, and you’re not just protecting your job from the agent sitting next to it. You’re building the resume for the role that manages the whole fleet.
Disclosure: This is not trading or investment advice. Always do your research before buying any cryptocurrency or investing in any services. Follow us on X @nulltxnews
Be the first to comment