https://agentsandhumans.ai/notes/doordash-s-reviewer-approved-35-000-changes-without-reading-them Agents & Humansen 2026-10-23T09:02:09.094Z DoorDash's reviewer approved 35,000 changes without reading them https://agentsandhumans.ai/notes/30-audited-platforms-let-humans-grant-access-but-never-say-what-stays-private Agents & Humansen 2026-10-22T09:02:14.553Z 30 audited platforms ask what an agent can do, not what it shares https://agentsandhumans.ai/note/arxiv-paper-shows-agent-groups-copy-each-other-into-bad-answers Agents & Humansen 2026-10-23T08:17:07.606Z arXiv paper shows agent groups copy each other into bad answers https://agentsandhumans.ai/note/doordash-ran-35-000-code-changes-through-a-multi-agent-system-for-4-79-each Agents & Humansen 2026-10-23T08:20:31.462Z DoorDash ran 35,000 code changes through a multi-agent system for $4.79 each https://agentsandhumans.ai/note/open-index-maps-how-much-of-a-job-agents-can-take-over Agents & Humansen 2026-10-23T08:23:32.517Z Open index maps how much of a job agents can take over https://agentsandhumans.ai/note/atlas-gives-designers-a-map-of-what-a-long-agent-run-actually-did Agents & Humansen 2026-10-23T08:32:06.546Z Atlas gives designers a map of what a long agent run actually did https://agentsandhumans.ai/note/abra-gives-each-agent-its-own-reusable-screen-panel Agents & Humansen 2026-10-23T08:18:58.360Z ABRA gives each agent its own reusable screen panel https://agentsandhumans.ai/note/hci-researchers-map-how-shared-workspaces-change-agent-work Agents & Humansen 2026-10-23T08:33:41.653Z HCI researchers map how shared workspaces change agent work https://agentsandhumans.ai/note/rct-finds-pausing-to-reflect-lifts-ai-assisted-writing-quality Agents & Humansen 2026-10-23T08:21:56.442Z RCT finds pausing to reflect lifts AI-assisted writing quality https://agentsandhumans.ai/note/a-new-study-finds-more-agents-hurt-quality-in-live-chat Agents & Humansen 2026-10-22T08:16:58.735Z A new study finds more agents hurt quality in live chat https://agentsandhumans.ai/note/arxiv-paper-makes-agent-promises-impossible-to-break-at-run-time Agents & Humansen 2026-10-22T08:18:28.996Z arXiv paper locks AI agents to rules they cannot skip at run time https://agentsandhumans.ai/note/30-platforms-audited-few-let-you-say-what-agents-may-share Agents & Humansen 2026-10-22T08:02:02.939Z 30 platforms audited, few let you say what agents may share