
LEGO-Anything shows promise, but benchmark results expose weak geometric self-assessment.
OpenAI introduced Dots at DevDay, positioning the GPT-6 Astra-powered assistant as an always-on agent that can pursue user-defined goals across tools and platforms with minimal oversight.

OpenAI announced Dots at its DevDay event, describing the product as a new personal agentic assistant powered by GPT-6 Astra. The company frames Dots as “remarkably capable, always-on agents built to handle everything.” The key shift is that Dots are designed to keep pursuing user-defined goals in the background with minimal oversight.
Unlike Codex or ChatGPT, Dots are meant to operate independently of a specific hardware device or interface. Users can launch Dots from Codex or ChatGPT, but the concept centers on ongoing work rather than single-session assistance. TechCrunch notes that much of the functionality was already possible through Codex and similar agentic harnesses, but Dots repackages it with a focus on independent agentic actions.
Dots are available starting Tuesday in ChatGPT for Pro and Business Premium users in eligible markets. Users can message Dots through Slack, Teams, and other organizational platforms, with text message support coming soon. OpenAI says users can begin with a primary dot, name it, and customize it, with a longer-term vision of teams of Dots working together on a user’s behalf.
OpenAI described examples including a developer using a dot to monitor customer feedback and implement fixes or requested features, and a scientist using a dot to rerun analysis as new experimental data arrives. The company also envisions “specialist Dots” with specific responsibilities, identities, credentials, and tools provisioned through existing systems. OpenAI is working with Microsoft to integrate Dots into Agent 365 security controls, a detail enterprise users should watch closely as always-on agents gain more access.
Dots also introduces a more cartoonish, approachable persona for OpenAI’s software. TechCrunch compares the branding to Meta’s Muse agent and notes that the dots are part of a broader push to make AI feel more relatable. For users, the practical takeaway is simple: Dots are not just a new interface, but a packaging of persistent AI work into a more visible, personalized form.

LEGO-Anything shows promise, but benchmark results expose weak geometric self-assessment.
Altman says AI benefits outweigh some harms, while rejecting catastrophic risks.
A former OpenAI safety leader says the company is not being careful enough.

RRSI aims to help self-improving AI agents generalize beyond their test tasks while using fewer tokens.