Tags
In an Anthropic toy run, one lead agent split Fermat’s Last Theorem among 999 helpers through a shared database, and they rebuilt the proof in 28.7 hours, against 11 days the first time. When you run parallel agents, do they share a task list or each get one job?
OpenAI’s dots gained a new skill, reading sign-up codes from your email to finish a 2FA check for you, though reports differ on whether they enter the code or pause first. Would you let an agent read the inbox where your 2FA codes arrive?
An alert fires, and an agent in Mistral Studio starts the incident investigation, follows the runbook and posts findings in Slack. Mistral says it saves hundreds of hours a month. Who starts incident triage on your team right now, an agent or a person?
