On May 5, Coinbase announced it would cut about 700 employees, roughly 14 percent of its workforce, as part of a restructuring positioned around AI. The same day, Bloomberg reported that PayPal will reduce its workforce by approximately 20 percent over the next two to three years. Meta is widely expected to announce another round of major cuts in the coming weeks.

This is the beginning of a shift that the market is not yet pricing in. For years, automation was framed as a gradual productivity gain. What we are seeing now is managerial. Companies are not just adding AI tools. They are redesigning the firm around the assumption that fewer people can do more work if agents sit inside the workflow.

The most important labor market change may not show up first in headline unemployment. It will show up in the falling unit price of knowledge work. As firms adopt agents, more roles get decomposed into tasks, reviews, and exception handling. That creates pressure on wages before it creates pressure on official job counts.

This is why Coinbase and PayPal matter together. They are early signs that companies are starting to treat AI as both a productivity layer and a workforce design tool. Once that logic spreads, the structure of white-collar work changes faster than labor data can capture.

Once tasks can be broken apart, routed, and benchmarked against machine output, companies stop paying for the full role and start paying for the minimum human intervention required. That pushes labor toward the lowest price at which the task can still be completed with acceptable quality. The mental model shifts. Companies stop managing people. They manage work.

On May 1, the Trump administration notified Congress that hostilities with Iran had been terminated for War Powers purposes. By May 5, Defense Secretary Pete Hegseth said the ceasefire was still in place despite new tensions around the Strait of Hormuz. On May 6 and 7, Iran was reviewing a U.S. peace proposal while tightening control over shipping through the Strait. Trump ordered U.S. warships to escort merchant vessels through the waterway, then abruptly shelved "Project Freedom" on Tuesday. Iran targeted oil facilities in the United Arab Emirates in the same window.

This looks less like peace and more like a frozen conflict.

Frozen conflicts create the illusion of stability without removing the underlying source of risk. Fighting slows. Diplomacy resumes. Markets calm down. The core dispute stays alive in a different form.

The Trump administration has shifted its approach. Reopen the Strait of Hormuz at any cost, and defer the harder negotiations on Tehran's nuclear and missile programs until later. That sequence has its own problems. The Strait has become a more immediate source of pressure for Iran than the nuclear file ever was, and that pressure does not expire when missile fire stops. The administration appears to have understood this only after it became expensive. It was obvious from the start. The likely outcome is managed tension.

Salesforce’s Headless 360 launch matters because it is more than a product update. It is an admission that the old SaaS interface is becoming less central. The browser is no longer the default point of control. The real contest is shifting to who owns the data, workflows, permissions, and trust layer that agents can act on.

What Salesforce is doing is straightforward but important. It is exposing the platform as APIs, MCP tools, and CLI commands so agents can operate across the full system without depending on a human clicking through the UI. In practical terms, Salesforce is trying to turn itself from an application into infrastructure.

That is the more important signal. Incumbent software companies now understand that their survival depends on becoming agent-ready before agents make their interfaces less relevant. If AI can reason, execute, and move across tools, the value does not sit in the screen anymore. It sits in the system of record and the accumulated business logic underneath it.

I think this is where the SaaS conversation starts to get more serious. For years, enterprise software was valued around seats, interfaces, and workflow lock-in. In an agent-driven environment, those moats weaken unless the platform can become the operating layer for machine action.

That is why Salesforce’s move matters beyond CRM. This is a preview of how large enterprise platforms will defend themselves against foundation models and coding agents. They will not win by preserving the old interface. They will win, if they do, by exposing their institutional memory in a form agents can use safely.

The second-order effect is a business model shift. Once agents become the users, per-seat pricing starts to look less natural. Consumption, orchestration, governance, and trust become more valuable than licenses tied to human logins. That is a real change in how enterprise software gets built, priced, and defended.