OpenAI’s Dev Day announcements left me disappointed. I can see the appeal of a personal assistant that keeps working between conversations, and I want to use better tools as they become available. But the promise of doing more arrived alongside an announced reduction in the Pro 200 allowance and another push to move more of the work into the cloud. For the way we have built our system, that combination is difficult to get excited about.
OpenAI introduced dots on September 29: persistent assistants with their own cloud computers. You can talk to them in ChatGPT and connect them to other services. Slack and Microsoft Teams are supported ways to reach them, and access to your own computer is an option too. There is room for more than one way of working here. The cloud computer, though, is part of the starting point.
We have built a self-hosted system over time. Our conversations and code live in tools we run ourselves, with choices we have made as our needs have changed. I want new assistants to be useful within that system. When the easiest connection assumes a different chat service, or a different home for our code, adopting a feature starts to involve decisions about the rest of the business.
The GitHub connection is most explicit in Codex Cloud’s setup, which starts by asking you to choose GitHub repositories. I can see why OpenAI would start there. Separately, Slack and Teams give people who already use them optional ways to reach a dot. For those people, a personal assistant can meet them in the software where they spend their day.
I feel the gap more because our choices have been different. The workplace connections in these announcements fit tools many companies already use. Getting the same convenience within a self-hosted system can mean doing more of the connecting ourselves. That extra work affects whether a new feature is useful enough for us to adopt.
There is a good chance we will outgrow parts of our current setup. We may need a different chat service as we scale, or decide to return to GitHub. I can accept those changes when they fit the work. I am less comfortable making them because the next assistant works most conveniently somewhere else.
The pitch that work can continue while my laptop is closed also leaves me cold. Codex Cloud offers that, and I understand why someone would want it. A self-hosted system can keep running while a laptop is off too. I am interested in having work continue; I also care about where it continues and how much control we retain over it.
OpenAI does provide choices. Dots can connect to your own computer, and the company describes permission settings and approval rules. Those are useful controls. I still have to decide how much access and ongoing context I want an assistant on a provider’s servers to hold. Running some work locally does not settle that question for me.
I have the same hesitation whether the company asking for that trust is OpenAI, xAI, or Meta. I want the efficiency these tools can offer, but an assistant that stays involved in my work asks for a continuing relationship with the provider. I need to be comfortable with the access I give it and with how I would pull that access back. Self-hosting creates responsibilities of its own, and it cannot remove every outside dependency. It does give us more say over parts of the system we have spent time building.
The Pro change makes me reconsider the arrangement further. OpenAI says new Pro 200 subscriptions have a lower included allowance. Eligible existing subscribers keep their previous allowance through October 29, 2026, then move to the lower allowance at the same $200 monthly price. OpenAI’s explanation points to increasingly efficient models. I want to see what that means for the work I can finish within the plan.
For the first time, I am seriously thinking about using Claude first. I have not made that switch, and changing providers would bring its own questions about trust and limits. But the lower allowance, alongside a direction that feels less natural for the system we have built, has given me a reason to reconsider where I start my work.
I want to try the new capabilities against the way we already work before deciding what to move. We may grow into more hosted services. For now, I would like the next personal assistant to make our existing choices easier to keep.