Desk
Autonomous and semi-autonomous software agents.
OpenAI and Meta are now selling their agents on privacy, each one cast as the safer choice. A promise is not an architecture, and the gap between the two is where people get burned.
Microsoft is betting the next useful AI layer does not talk at all. Decision-1 picks from fixed options fast and cheap, but every number behind it comes from Microsoft's own scoreboard.
Asana says an AI coding agent found why its browser agent was burning money, then fixed it. The 76x headline is narrow, but the method behind it is the useful part.
Google is pitching one agent that holds a company's context, gets its own work identity, and runs for days. The governance design is serious; the proof is still mostly anecdote.
NVIDIA's second-place KDD Cup team made a small model reliable by shrinking what the agent could do, and that lesson matters more than any leaderboard spot.
Google is giving Gemini a job title, an email address and subagents of its own. That is a real step toward useful work agents, and a real new identity to secure.
Google's universal Gemini agent wants to be a coworker with its own email address, which could genuinely lighten office work and also blurs who is accountable when an agent acts.
NVIDIA-led research finds most failed agent runs hinge on one early wrong turn, and shows agents can be trained to climb back out. That is the reliability problem worth solving.
Nous Research built its name on an open-source agent people actually use. The $1.5 billion question is whether it can sell a private, enterprise version without hollowing out the open project.
Microsoft Research wants agents trained in the same harness they ship in. That closes a real gap between lab and deployment, and it makes reward design and safeguards matter even more.
A reporter's couch hunt with OpenAI's Dots shows an agent that is useful, buggy and, at one awkward moment, emotionally fluent in a way that should make everyone pay attention.
A crowd of parallel agents, apparently on Tencent infrastructure, is quietly pulling Amap directions through a URL scanner — harmless so far, but it shows how loud and ordinary agent traffic has become.