I'd love for Dot to support multiple Dots, like Grok Bot does, so I can keep different work in separate conversations and avoid mixing up context. @thsottiaux
Outside of work, Ning and I built a little Skill together. ๐ฑโจ
When you're reading a paper closely, flipping back and forth between the text and the figures gets old fast. So we put them on one page: text on the left, figures on the right. As you read, the right side switches to the figure you're on, so no more flipping around ๐
Give a paper's PDF to your agent, and the micro-paper Skill turns it into a page like this. Next up, we want to hook it into Zotero so your papers get filed there too.
OpenAI Dots is starting to look less like an assistant and more like an actual employee
Peter Yang gave it real work from his one-person business for a few days:
it found a five-figure sponsorship issue, analyzed 1,800+ YouTube comments, helped launch his app and even built a paid product
18 minutes of real Dots workflows:
00:42 โ how Dots actually works inside ChatGPT
03:18 โ researching X + Slack and creating content
07:38 โ analyzing 1,800 YouTube comments across hundreds of videos
08:29 โ finding app bugs, creating App Store assets and auditing sponsors
11:26 โ turning an idea into a paid product + landing page
13:29 โ where Dots still breaks and how it compares to Grok Bot
probably one of the better real-world demos of OpenAI Dots so far