🤖 At DevDay, OpenAI introduced dots - background AI agents that operate in ChatGPT not within a dialogue, but continuously, between conversations.
The idea is that a dot doesn't wait for your message but continues the task on its own: it has its own cloud computer, access to 4000+ applications, and memory of what you've asked before. You can assign it to monitor user feedback and fix bugs, recheck experiments with unexpected results, and work through Slack and Teams (SMS promised later). The first dot can be named and customized.
OpenAI further promises teams of dots that work together, and "specialized" dots with separate access, corporate hardware, and integration into company systems - for this, there is already a partnership with Microsoft and Agent 365.
Technically, this is powered by GPT-6 Astra, available on Pro and Business Premium plans, with the first dot included for free. Bloomberg reports on a new plan around $500 per month for more serious use.
Interestingly, shortly before this, Meta showcased Muse with the exact same idea of a cartoonish agent-avatar with a name and character - it seems this is now the industry's common language for selling autonomous agents to the mass user. Meanwhile, the dot itself gains access to thousands of applications and operates with minimal human control - and this is no longer a UX issue, but a matter of access rights and who is responsible if the agent makes a mistake.
https://openai.com/index/introducing-dots/
#openai@rvnikita_blog #ai@rvnikita_blog #agents@rvnikita_blog #chatgpt@rvnikita_blog