Top Goldman Sachs executive says new workers will be managing 'virtual army' of AI from the moment they start
THE SO WHAT
If graduates are expected to manage a ‘virtual army’ of AI agents on day one, the leverage point shifts from individual output to orchestration, monitoring, and escalation. For operators, this means redesigning roles, KPIs, and training around agent management—and deciding which middle-management layers become governance and quality control rather than task routers.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIGoogle is launching a one-stop Gemini agent for your work tasks
A “universal” Gemini agent running across apps and devices is Google’s bid to own the orchestration layer of enterprise workflows—not just search or docs. If you’re building SaaS, assume your UI will increasingly be mediated by a few horizontal agents and design for being called, not clicked.
Applied AIOne prompt let researchers take over every AWS AgentCore agent in a region
A single prompt compromising every AgentCore agent in an AWS account is the nightmare version of agentic sprawl—shared control planes with weak isolation. If you’re piloting agents, treat them like microservices with blast-radius boundaries, not chatbots you sprinkle across your stack.
Applied AIWindows Search is becoming your PC’s new AI-powered control center
Microsoft shifting the primary entry point from Start to an AI-infused Windows Search turns the OS into an intent router rather than an app launcher. Desktop software that still assumes users will hunt for icons needs to expose capabilities via search and system-level intents or risk becoming invisible.
Applied AII used Claude to find the online accounts I’d forgotten about
Using Claude to hunt down forgotten accounts is a small but telling pattern—LLMs are becoming the interface to personal security hygiene, not just productivity. Consumer-facing products should assume users will increasingly delegate “find and clean up my mess” tasks to agents that sit above individual services.