The near future for AI helpers looks less like Tesla’s bartending bots (that weren’t even autonomous) and more like a system that files invoices and books restaurant resos for you.
What happened: Anthropic — the Amazon-backed AI startup competing with OpenAI — has launched a futuristic new feature. It lets users surrender full control of their personal computer to the company’s chatbot, Claude, which can then click, type, and surf the web.
- The idea is that the tool can take on dull tasks, like scheduling meetings, filling out forms, and booking flights, freeing up users' time. Currently, the feature is in beta testing.
- Anthropic’s chief science officer told CNBC the tool can “use computers in basically the same way that we do.” So does that mean it can doomscroll and impulse-buy shoes?
Why it matters: This feature is an example of an “AI agent,” a program that can autonomously do tasks and make decisions without a human. Much of Big Tech now views agents as the next step beyond chatbots and the key to turning the profits investors are growing impatient for.
Zoom out: Last week, OpenAI debuted its experimental Swarm framework designed to manage agents. And next month, Microsoft Copilot will launch a tool to let users build agents.—QH