The Bot Is Running. Your Job Is to Watch.

Walk through a Bay Area office on any given morning and you might notice something unfamiliar: a knowledge worker at their desk, coffee in hand, watching a browser tab autonomously navigate web pages, pull data, and assemble a summary. They're not clicking. The bot is.

That's the tableau the Wall Street Journal documented this week in a feature on AI agents and the Silicon Valley professionals increasingly relying on them. Engineers, executives, and knowledge workers are handing off the procedural layer of their jobs — research, data entry, web navigation, email drafting, code generation — to autonomous systems built on large language models, then watching those systems execute.

Anthropicʼs Claude features centrally in the piece. Its computer-use capabilities and tool-calling infrastructure allow the model to operate within live software environments — opening browsers, running scripts, interacting with applications — and the Journal cites it as one of the more mature implementations of this approach on the market. The framing is competitive rather than singular: multiple vendors are building toward the same workflow, with different technical bets on how much autonomy to hand the model and how to keep humans appropriately in the loop.

The human role the article describes is a studied one. The worker doesnʼt disappear — they shift. From executor to approver, from doer to supervisor. 'Bot-watching' has become a recognizable description of how some people spend portions of their workday, and enterprise software companies are adjusting their thinking accordingly. A product whose primary user is monitoring rather than operating is a different design problem.

What the Journalʼs attention signals is less about technology and more about perception. Autonomous AI workflow has been a developer conversation for years. It is now a business-press story — which means the population of people expected to have an opinion on it has expanded considerably. Whether that translates into durable enterprise adoption beyond a technically fluent early-adopter cohort, or stalls as a high-visibility novelty, is a question the market hasnʼt settled.