The latest OpenAI release has caught my attention for the way the app brings the work together. Models are part of that, of course, but the working environment is what has me changing where I spend my day.
OpenAI has just released GPT-6.1 Sol. On paper, it looks capable. New models arrive so frequently now, though, that another benchmark chart rarely tells me much about whether a tool will become part of my daily work. I suspect most of us still have plenty of room to get more out of the models we already have.
What interests me more is how easily I can put that capability to use, and whether the context, documents and tools I need are there when I come back tomorrow.
The work has been spread across too many tools
I’ve spent a lot of time using Gemini, Claude, Cursor and other AI tools. Each has things I like. I’ve enjoyed Claude’s models, even though the way the app organises work has never quite suited me. Cursor has been useful for coding. I’ve also been using Grokbot over the past month or two, and it has been a massive help with getting work done.
For a long time, I’ve maintained my own context in a provider-agnostic setup. Instructions, working knowledge and reusable skills need to be useful across applications. I want to be able to use the tool that suits the task without rebuilding everything I’ve already taught another one.
That has worked, but keeping the pieces together takes effort. The model can be excellent and I can still spend too much time finding the right conversation, moving material between applications or explaining context again.
Running ChatGPT on macOS now feels closer to the cohesive tool I’ve been hoping for. I can see myself doing a much larger proportion of my work in one place.
Space and pages make a practical difference
The features I think deserve more attention are Space and pages.
We’ve had folders, projects and files supplying context for a while. An editable page gives that work somewhere to live as a document: notes, a brief, a research summary or supporting material that I can return to and keep working on.
Pages can also contain charts, trackers and interactive tools. That makes the document more useful than a static record of a conversation. Being able to generate something I can interact with inside the material I’m already working on is a particularly clever part of the experience.
I like having those pieces close together. The plugin integrations and access to coding work make the whole application feel more useful to me, because there are fewer decisions about where the next part of a task should happen.
I want to try dots, but portability still matters
I haven’t had access to dots yet, so I can’t say how well they work in practice. OpenAI describes them as always-on agents with their own cloud computers, able to work towards goals over time and connect to other applications through plugins.
That is an appealing prospect alongside the context and work I’m already managing. If the experience holds up, it could cover some of the work I currently give to Grokbot.
My hesitation with Grokbot is the amount of control I have over what lives inside it. I don’t have the straightforward access I’d like to a folder of context, or a simple way to move agent rules, skills and scheduled tasks into another application.
I understand the appeal of a managed experience. Plenty of people want the agent to work without having to organise the underlying files themselves. But once I’ve invested time in teaching a system how I work, being able to retrieve and reuse that knowledge matters to me.
I’ll be looking at dots with the same questions. Can I understand and manage the instructions it follows? Can I retrieve the context I’ve built? How much of the setup can I carry across if I choose another tool?
Bringing work into one application is convenient. I still want to retain control over the material that makes it useful.
Why I’m making the move now
For now, the overall ChatGPT experience has become good enough that I’m moving the majority of my work there. Space, pages, plugins and the connection to coding work are the reasons I’m spending more time in it.
That is also how I tend to think about AI in our work at Rabbit Hole Digital. Model capability matters, but so does how it fits into an actual process: where the information lives, how the tools connect and how much control the person doing the work retains.
I’m looking forward to trying dots. In the meantime, the application already feels much closer to the working environment I’ve wanted.
Sources for product details: GPT-6.1 Sol, Space and pages, Introducing dots.
