OpenAI, Anthropic and Microsoft each launched a dedicated AI "coworker" product within a five-month stretch, and the corporate job cuts came almost immediately after. OpenAI shipped ChatGPT Work on July 9, built on its new GPT-5.6 model, bundling its Codex coding tool and a persistent cloud agent into a single desktop app. Anthropic beat them to general availability: Claude Cowork went live in April and now appears as a third tab in every paid Claude plan, sitting alongside Chat and Code, reading files and running scheduled tasks autonomously. Microsoft's Copilot Cowork hit general availability in mid-June, wired directly into Word, Excel, Outlook and Teams, and it is already deployed inside more than half of the Fortune 500.
What these tools actually did at companies that tested them
Microsoft documented one engineering team using Cowork to compare nearly 4,000 files across two product versions, work that would have taken weeks compressed into hours. A sales team fed it a stalled pipeline and got back a ranked list of dead deals, each one paired with a ready-to-send follow-up message, collapsing a week of manual review into a single morning. Salesforce reported its AI agents now handle roughly half of all customer support interactions. These are measured outcomes from real deployments, not press release projections.
The job cuts followed the deployments with little delay. Salesforce cut 4,000 support roles after rolling out that same agent technology, then trimmed another 1,000 this year. IBM replaced close to 200 HR positions with agents while tripling entry-level AI hiring. Amazon has shed roughly 30,000 corporate roles since late 2025. Across the tech sector, layoffs are running at more than 1,100 per day in 2026, nearly double last year's pace, and companies are increasingly willing to name AI as the direct cause, something that would have been a PR liability eighteen months ago.
The data is murky, and that suits employers just fine
Pinning exact numbers on AI-driven displacement is genuinely difficult. Challenger, Gray and Christmas tracks layoff announcements, but companies rarely break out which cuts are AI-related versus restructuring or revenue pressure. That ambiguity is not neutral: workers trying to assess their own exposure get "AI washing" instead of straight answers, a fog of strategic vagueness that benefits the companies doing the cutting. The $700 billion infrastructure buildout funding these tools flows directly from payroll savings that nobody is required to itemize.
For workers still in the room, the practical pivot is from doing tasks to supervising agents doing tasks. The role that survives is closer to a fleet manager than an individual contributor: setting goals, checking outputs, catching errors that no agent flags itself. That shift is already happening faster than most job descriptions have caught up to it.
Tech stocks ticked higher on the week these launches were announced, with Microsoft shares up roughly 2% on Copilot Cowork's general availability date.
This article is for informational purposes only and does not constitute financial or career advice.
