My Thoughts on Dot After 7 Days

I've been using Dot for the past week straight, and the more I use it, the better it gets.

Dots are OpenAI's always-on agents inside ChatGPT. Each one has its own cloud computer, connects to your apps, and keeps working on things between conversations. You can message it or call it.

My primary use case is voice. Most of what follows comes back to talking to it, while it has access to everything I've connected.

The word dots glowing over a dark studio — Dots launched on September 29. OpenAI

Why I like Dot more than ChatGPT or Codex

One main thread is way better

My biggest pet peeve with ChatGPT and Codex is searching for chats. Every task lives in its own conversation, and I'm always hunting for the right one.

Dot has one main thread. I give it a task and forget about it. It comes back to me in the same place.

It has an identity

It's one agent with a name and a face, not a fresh chat window every time. That sounds small, but it changes how much I'm willing to hand off.

A Dot named Felipe working through a slide update in its one main thread — One thread, one agent. It works next to you and reports back in the same place. OpenAI

It writes better Codex prompts than I do

Dot can write the prompt and send it to Codex for me. It does a better job than I would, especially on my phone, where typing a long, careful prompt is the last thing I want to do.