Giving an artificial intelligence software direct control over an isolated virtual computer sounds like an engineering nightmare or a massive leap forward, depending on who you ask over coffee. OpenAI isn’t playing it subtle anymore.
The company just pulled back the curtain on “dots”—always-on autonomous AI agents powered by a new GPT-6 Astra foundation model.
These digital entities are built to live inside your workplace systems, monitor your communication channels, and execute complex work without waiting for you to type out a prompt every five minutes.
Key Takeaways
- Autonomous Cloud Sandbox: Each dot operates on its own dedicated cloud computer and browser environment, allowing it to code, analyze data, and run background tasks around the clock.
- Deep App Integration: Powered by GPT-6 Astra, dots connect to over 4,000 apps via plugins, operating fluidly across ChatGPT, Slack, and Microsoft Teams.
- Enterprise Governance Play: OpenAI is partnering with Microsoft to bring “specialist dots” directly into Microsoft Agent 365 for administrative, IT, and compliance workflows.
Instead of acting like a basic chat box sitting in a browser tab waiting for instructions, a dot behaves more like a junior employee with its own dedicated hardware.
It monitors incoming messages, flags problems, and attempts to solve them in the background.
If a software bug pops up in a dedicated engineering channel, the agent spins up a sandbox, investigates the code, writes a patch, and presents a complete pull request for human approval.
OpenAI outlined the fundamental mechanics behind the launch in its official announcement:
“Dots are frontier intelligence that have your back. Powered by GPT‑6 Astra, they have their own cloud computer, learn from feedback over time, and can work towards your goals 24/7. Through our ecosystem of plugins, they can readily connect to over 4,000 apps, giving them the tools to help wherever you need them.”
That 24/7 background operation hinges on a concept OpenAI calls “proactive research.”
When you aren’t actively messaging your assigned agent, it scours your connected workplace tools looking for open tasks.
To keep things from going completely off the rails, these background checks operate under strict read-only permissions.
The agent can’t independently fire off emails, modify databases, or tweak system configurations without explicit human sign-off.
Cloud Hardware and the GPT-6 Astra Engine
What sets this release apart from previous AI assistant launches is the sheer amount of infrastructure handed directly to the model. Each dot runs on its own cloud-hosted computer, complete with an isolated browser and app execution layer.
If you want to watch the machine work, you can open up a virtual window and inspect its screen in real time as it clicks through web pages or writes scripts in Codex.
For users who need local execution, you can also grant a dot permission to link directly to your physical laptop. That allows the agent to manipulate files on your desktop while you grab lunch or handle client meetings.
You can communicate with your assigned dot across multiple touchpoints—whether that’s dropping a text in Slack, initiating a project inside Microsoft Teams, or jumping on an ambient voice call through the ChatGPT mobile app.
The company is pitching specialized use cases across several distinct professions:
- Software Engineering: Agents scan customer issue threads, draft bug fixes, run automated tests, and submit review-ready pull requests paired with video walkthroughs.
- Enterprise Sales: Dots evaluate prospect account histories against product documentation, build custom technical proofs-of-concept, and update sales proposals as requirements shift.
- Product Management: As project scope shifts, agents automatically revise pitch decks, recalculate timelines, and draft updated launch documentation across connected drives.
- Scientific Research: The system continuously reruns analytical models as new raw datasets drop, updating charts and flagging statistical anomalies for human review.
Governance, Auto-Reviews, and the Microsoft Partnership
Handing background access to corporate data raises obvious security flags, especially for risk-averse enterprise IT departments. To counter those concerns, OpenAI built an automated action-review framework.
The software evaluates potential actions against preset Custom Rules, automatically blocking high-risk maneuvers—like changing account passwords or exporting sensitive user databases—and routing them straight to human managers for manual approval.
Furthermore, OpenAI is aggressively targeting the corporate IT stack by partnering directly with Microsoft.
The two companies are working to plug “specialist dots“—agents built with unique organizational identities and system credentials—straight into Microsoft Agent 365.
This integration will let enterprise administrators provision, manage, and audit autonomous agents using the exact same compliance dashboard tools they already use for human employees.
Rollouts are starting immediately for subscribers on ChatGPT Pro and Business Premium tiers in eligible markets, with Enterprise and Edu account administrators able to toggle on a beta the day of launch.
Your primary dot comes bundled into standard plan pricing at no extra base charge, including a generous allowance for deep computational tasks during the initial launch month.
We’ve been promised “agents that do our work” for years now, mostly resulting in clunky wrappers that break the moment a UI changes. But pairing a GPT-6 Astra brain with a dedicated, cloud-hosted virtual machine changes the math significantly.
Whether companies are actually ready to manage fleets of digital workers running on virtual computers remains an entirely open question, but the tools are officially sitting on the desk.
Discover more from BITVoxy Global
Subscribe to get the latest posts sent to your email.
