OpenAI Dots after one week: attention-saving agents, but permissions still frustrating

danshipper · x · 2026-10-08

Every's Dan Shipper reviews OpenAI Dots after a week of real use:

Attention protection: his Dot posted model-testing results to Slack and brought back replies, so he never fell down a Slack rabbit hole.
Catching small stuff: extracting school-email key points, surfacing missed Slack messages, flagging account-access requests.
Voice on the go: the head of video directed edits via voice while away from his laptop.
Ecosystem advantage: roughly feature parity with Muse, Instinct, and Grok bots — but integration with your existing ecosystem matters.
Pain point: approving the same basic permission multiple times, and agents sometimes claiming they can't act even after approval.
He doubts loyalty to agent personas; expects a switch to persistent agents within a year.

The accompanying podcast also covers personal vs. company agents and whether firms will build a 'shadow org chart' of agents.

More from coding & agent