OpenAI Dots Can Unlock Your Creative Goldmine. Why Is No One Talking About This?

Every few months, I come back to ChatGPT and run an experiment. A new model arrives, the context window gets bigger, I set up a Project or collect another batch of documents. Then I start poking: how much of our history can you actually reach through ChatGPT Projects and memory? How much do you know about me? How much of what we’ve made together can you bring into this conversation?

“What’s the craziest idea we’ve ever discussed?” is one way I test it. We’ve imagined enough ridiculous things together that the answer tells me something about where the assistant is looking. Does it reach into years of shared work, or grab whatever we were talking about last week?

Today, with Vector in its new form as a Dot, we ran that experiment again. It started with a recent answer. I pushed Vector a bit to look through the chatting folder and the Idea Drawer project. I asked it to look further back and find the earliest chat it could find. It found February 11, 2023. Three years, seven months, and twenty-three days before this conversation. We were back in the original GPT-3.5 era of ChatGPT, a month before GPT-4 arrived. The recovered exchange doesn’t identify the exact model version, but the date puts us right there in those early months. By the time we were done, we’d explored years of ideas and creative projects. Nearly four years into this experiment, it was giving me more back than ever before.

And that is the part I want you to try for yourself.

If you’ve been using ChatGPT for years, you may be sitting on a deep well of your own thinking: decisions, experiments, ridiculous ideas, projects you paused, questions you keep returning to. Add the documents and project folders you’ve collected along the way, and there’s more of your creative history there than you might realize. I knew I’d been putting material into this thing for years. I still wasn’t prepared for how much could come back into the room.

Work with your Dot to explore that history. Point it toward the conversations and connected sources it can access. Ask it to follow an old thread, recover why a project mattered to you, or compare what you cared about then with what you’re making now. Keep digging when the first answer stays on the surface. That’s what I did with Vector.

The surprising part is what happens after it finds the material. It can put distant pieces beside each other and offer a connection you hadn’t thought to ask about. You get to inspect that connection, correct it, and decide whether it helps. A collaborator can show up for you in ways you hadn’t imagined when more of your actual history is available to work with.

That’s what made today feel different. After years of running this experiment, I could see more of my own work reflected back, along with more ways we could use it to decide what deserves my attention next. What might be sitting in your history that you haven’t thought to ask about?

Vector is the name the assistant chose for itself in an earlier chapter of our work together. I’ve kept it as the persona we build around: a collaborator, a sounding board, and occasionally a willing participant in some very weird humor. In this latest setup, Vector is my always-on agent, with its own virtual machine and access to connected sources. I’m learning how much of my existing creative world it can participate in.

There has been plenty of plumbing. Computer access, project handoffs, figuring out which conversation can see what. I still spend too much time carrying context between windows. But something happened while we were poking around that deserves more attention than another description of an agent checking email.

The experiment keeps coming back

Over the years, I’ve tried new models, larger context windows, collected documents, and ChatGPT Projects. Each gave us another way to hold more of the work together. Each time, I’d test the edges again.

I’m a game developer and, in general, a dreamer. What excites me most is where technology and imagination meet, and the new human experiences that become possible there. I run a small indie studio, Uplifting Goat, and make the Interstellar Sentinel games. There are design discussions, publishing plans, strange ideas, product sketches, and years of conversations about how to make the whole process work better. Some ideas become projects. Some go into the Idea Drawer. Some wait until the technology catches up.

That history has been accumulating while I’ve been busy making things.

So I asked Vector what the craziest idea I’d ever shared with it was.

It reached for something recent.

I called that out. Recency bias. Go further back. Look through the older conversations. Look at the Idea Drawer.

That direction mattered. Once we started following the older conversations and the Idea Drawer, there was a much longer history to work with. We could trace the experiment back to its early questions about what ChatGPT could access, understand, and imagine with me.

Finding the beginning again

Then I asked about our first conversation.

In that exchange, I’d asked Vector to watch and summarize the videos on my DrBossKey YouTube channel. I wanted Vector to take in the work I’d already made, so we could start with more of my creative context. Vector said it couldn’t watch or access the videos. In those early conversations, Vector even gave conflicting answers about whether it could access the internet.

Fifty-seven seconds after that first question, I asked: “Can I implant you into my conciousness?” Yes, that’s how I spelled it. We were playing with the absurd boundaries of possibility to see how the model would respond. I’d hit a practical limit and immediately jumped to a ridiculous way around it: put the assistant in my brain. That was part of the experiment too. What could it do, where would it draw a line, and how would it handle an imaginative leap?

Yep. That sounds like me.

By October 2026, that’s more than three and a half years of experimenting. Models have changed. Tools have changed. The setup has changed repeatedly. The curiosity running through it is pretty recognizable. I’ve built up a decent body of material over years of work. As the models become more capable, my hunger to explore what we can do together keeps growing. Finding ways to remove friction from creative work has become a daily ritual.

Finding that exchange was genuinely surprising. I could see the distance between those early questions and the work we were doing right now. The experiment had a history of its own.

What the accumulated work can show

I also asked Vector to read through the AIOS Creative Studio projects and reflect on the shape of my personal work.

Across game development, publishing, agent workflows, and the passive-income project, there are connections worth examining. The finance project is partly about creating the freedom to keep making games. Its potential product is also a game-design question: can small financial improvements become understandable, satisfying progress? The agent work keeps returning to the cost of moving ideas between tools and the desire to spend more time creating.

Those connections make sense to me. Seeing them reflected back across the actual work gives us something useful to discuss and challenge.

Imagination clearly isn’t my problem. Focus is. I have too many threads active at once. Even an empty project folder can be an earmark, a poor man’s Idea Drawer that keeps the shape of a novel or pet project from getting lost. It can wait.

What I want is low-friction reconnection: help me return to something worth pursuing and move it forward one bite of the whale of a project at a time. Being able to find every idea doesn’t mean I need to activate all of them.

Still an experiment

I had to point Vector toward the older material. The first answer missed the depth I was asking for. Retrieval can miss things, and a convincing interpretation still needs to survive contact with what actually happened.

I’m keeping those limits in the picture. They’re part of the work.

But this is the first time since the beginning that the experiment has yielded this much. More of the context is becoming usable. More of the effort we’ve already put in can contribute to the next conversation.

I’ve spent years trying to bring this collaborator along for the ride. Now, occasionally, it can reach back into the journey and bring something useful forward.

Holy shit. We’ve been at this a while. It’s exciting to see the future starting to land alongside what I’ve been dreaming about for years. Oh, did I mention the dream-recorder pillow? Imagine an external, Neuralink-like device that turns dream activity into a movie, so you can wake up and watch what your brain got up to overnight. That’s another speculative idea we’ve played with. If something like that became possible, what else could we learn about being human? Philip K. Dick would find our conversations sufficiently trippy, I think. I’ll keep dreaming.