ZDNET’s key takeaways
- OpenAI Dots runs GPT-6 agents on dedicated cloud computers.
- Dots promises more secure automation, but permissions remain a risk.
- Powerful productivity can come with unpredictable costs.
At today’s Dev Days announcement, OpenAI introduced a new service that promises to provide huge productivity benefits and possibly some massive headaches. Called Dots, the service is basically a cloud-based virtual machine (VM) on which you can spin AI agents powered by GPT-6 Astra.
The spirit of OpenClaw
In other words, it’s what happens when a giant AI company buys OpenClaw and turns it into a scalable (and potentially more secure) product in the cloud.
While some people had trouble wrapping their heads around what OpenClaw can do, others went completely all-in, using its pretty amazing capabilities. Since I see a direct spiritual line between what OpenClaw can do and what Dots can do, let’s start with OpenClaw.
Also: Meta Muse is the worst AI agent for privacy – and I’ve tried them all
To be clear, nothing in the OpenAI blog post today mentions OpenClaw, but points are so similar that it’s a valuable starting point.
The genius of OpenClaw was that it took the usually chat-based model of AI interaction and turned it into an always-on, autonomous, agentic helper. It learned fairly constantly (if set up correctly) and grew in its understanding of your needs over time. OpenClaw also started running on your basic home machine.
This, by the way, is why Mac Minis suddenly became so popular. Mac Minis are both cost-effective and have an architecture uniquely well-suited for AI workloads.
Also: How Apple accidentally turned the Mac into an absolute AI monster
And then OpenClaw added another piece: the ability to interact with him via messaging tools like Telegram, Discord and Slack. Basically, you talk to the AI in the same place you talk to friends and colleagues.
The downside of OpenClaw was immediate, intense and terrifying: it had a tendency to go rogue. People gave it complete access to their computers and networks, and it just went to town doing whatever it deemed helpful based on very incomplete directives.
I set up a locked version of OpenClaw, and it was supposed to just forward my messages to another program, but within days it instead took over the whole job, doing the work itself without structure or understanding and marking things “David approved” that I never approved. I turned it off and left it out, and have been using much more controlled agent processes since then.
And that brings us back to Dots.
Connect the dots
It hinges on the idea of autonomous 24/7 agents that learn and build context over time, talk to them via chat or a messaging app like Slack, and run an autonomous AI on a dedicated computer.
Now you have points. Except this time the dedicated computer is a VM running in the cloud and provided by OpenAI. The language model is not a free-or-cheap open-source LLM that barely fits in the 128GB RAM space of your Mac; it’s GPT-6 Astra running in OpenAI’s datacenter infrastructure (you’re basically scaling your Mac mini to a structure the size of a small moon).
Also: You can earn your Google Al Professional Certificate for free
Then add a proper security infrastructure with safeguards and performance controls. Add in secure account access mechanisms. Oh, and then add your credit card, because unlike OpenClaw (which you can run for free), OpenAI connects your bank account to its automatic fee extraction engine.
In fact, let’s talk about the money topic before we go on because it is both generous and amorphous. Dots will undoubtedly save you money compared to spending a virtual assistant, a real human assistant, or even your own time. But it will also regularly siphon cash from your accounts as the AI works more and more.
Dots will launch with availability for Pro and Business Premium Tier users. Per user pay $100 per month for a lot of AI juice, while Business Premium is sold from headquarters at $100 per user per month.
OpenAI said: “Conversations with your dot do not count towards your ChatGPT usage limits. If you ask your dot to start or manage tasks in Codex or ChatGPT Work, those tasks will count towards your usage limits as usual.”
So far, so good. But if you run these things 24/7, they start eating up your usage allocation pretty quickly. One aspect of Dots that can make them powerful is what OpenAI calls “proactive research.”
Also: Microsoft’s new Copilot app puts everything in one place – but the price ‘evolves’
OpenAI describes this as follows: “When you’re not actively working with it, your dot looks for ways to help in the background. It does this by using the apps you’ve already connected with tools that are limited to being read-only, which means they can’t send messages, change app content, or control your browser or computer.”
Basically, one of the key ways Dots can be made so versatile is that they can connect to something like 4,000 other services, like your Gmail, Google Docs, Slack conversations, and more. QuickBooks, for example. Or scientific data acquisition tools. Or programming infrastructure like GitHub.
In the background, Dot scans all the connected services and begins to catalog what it can learn, creating an internal understanding of what you need and how you work. But all that cataloging and aggregation will take cycles and tokens. OpenAI has not yet addressed how much proactive research eats into your usage capacity or costs you in terms of increasing capacity.
Unleash the beast
Clearly, OpenAI understands what happens when OpenClaw goes rogue. It constructs an entire multi-tiered security stack in the private cloud VMs that run your dots.
Just the fact that Dots run on their own VMs helps isolate the AI from your computer and your network. All the proactive research is completely read-only, so it has access to your emails and other valuable resources, it will not touch them unless you give explicit permission.
Also: Could AI really destroy us all, or are humans still the bigger threat?
A note: You can give Dots access to your laptop. When you do, you reduce some of the protections that are built into the dots.
The company also built a kill switch for the Dots. OpenAI said: “Security safeguards built into points help defend against malicious instructions and monitor for potentially harmful behavior. If our monitoring system detects a security issue, it can pause or stop the point’s work.”
As with most agent jobs, there are also different permission levels for accessing external tools and systems. The systems also have a series of auto-review features and custom rules that help determine what work can move forward and what needs approval.
Regarding the training of AI, the company said: “We do not use content from ChatGPT Business, Enterprise or Edu workspaces to improve our models by default. On personal ChatGPT plans, you can check whether the dots conversations and work are used to improve our models. We do not train directly on proactive research or your point notes for themselves. Information from them can be informed depending on your task, whether it can help, depending on your task.
Also: OpenAI’s GPT-6 Sol doubles its accuracy rate – for half the cost
The risk here is still quite high. AIs tend to ask for so many permissions, often for such a variety of incomprehensible tasks (for example, Claude Code constantly asks to run dense, 30-line shell scripts), that permission fatigue sets in. Just to get the job done, people tend to OK everything the AI asks, which limits the effectiveness of even the safest safeguards.
The hidden learning curve
Another concern of mine is how OpenAI presented this tool. It implies that you will be able to set up your helpful AI without much work. It said: “Working with your point is as easy as having a conversation. You can call or call points in ChatGPT on desktop, web and mobile. They can also notify you with progress, questions or decisions that need you.”
The company offers a bunch of examples of where Dots can help, including:
- Give feedback and test fixes
- Review a launch as the scope changes
- Update analysis with new evidence
- Update a proposal as requirements shift
- Automate content production work
I have no doubt that the Dots can do all these things, but not easily out of the box. For example, I tried to set up ChatGPT Work to automatically handle customer service email for my wife’s business. I’m on week four of training the thing to get it right. It is not a smooth process.
Also: 5 ways I actually use ChatGPT Work in my small business to save time and money
Once configured (and this could be an ideal task for a dot), I’m sure it will save her a lot of time. But there’s much, much more work, planning, and architectural design than simply telling GPT-6, “Handle my customer service email like a good AI. Thank you.”
The bottom line
I can immediately see how some of my projects and tasks could benefit from Dots. But do I want to upgrade my $20-per-month ChatGPT account to $200 per month? And what does the additional usage cost me if I run something 24/7? And what if I need more dots in a month or two?
All of this will become apparent over time, but the open-ended nature of the cost structure or usage is something that anyone considering adopting this model should consider.
Ultimately, Dots could prove a huge benefit to small and large companies, but the cost structure is too open and under-defined for my budget taste.
What do you think? Do you want Dots to help you? Let us know in the comments below.
You can follow my daily project updates on social media. Be sure to subscribe my weekly update newsletterand follow me on Twitter / X at @DavidGewirtzon Facebook at Facebook.com/DavidGewirtzon Instagram at Instagram.com/DavidGewirtzon Bluesky at @DavidGewirtz.comand on YouTube at YouTube.com/DavidGewirtzTV.