The Internet.
I’ve been thinking a lot about what the ideal AI interface would look like for someone with a neurodivergent mind, and the more I sit with it, the more obvious it feels: the interface I want already existed once. It lived in the terminal. It lived in IRC. It lived in HexChat. It lived in that era when computing was quiet, textual, predictable, and built around the idea that thinking should come before spectacle. Back when the loudest thing your computer did was beep because you forgot a semicolon.
For decades, the internet was a sanctuary for people who think the way I do. It was slow in the best way. It was patient. It was asynchronous. It let me process at my own pace. It let me organize my thoughts in parallel threads. It let me communicate without performing. Then RealPlayer arrived, and Flash after it, and suddenly the web wasn’t a reading space anymore. It became a broadcast medium. Autoplay, animation, video ads, motion everywhere — the sensory load skyrocketed. It was like going from a library to a Best Buy demo wall overnight. And if you were autistic, it felt like someone had replaced your quiet terminal with Clippy on a Red Bull bender.
AI chat interfaces have been the first major reversal of that trend. They brought back stillness. They brought back black‑screen/white‑text minimalism. They brought back the feeling of sitting in a quiet room with a single thread of thought. But even now, the interface is still built around one long conversation. One scroll. One context. That’s not how my mind works. I think in channels. I think in compartments. I think in parallel threads that don’t bleed into each other. And I think best in a terminal — a place where everything is text, everything is predictable, and nothing moves unless I explicitly tell it to, the way nature intended.
That’s why the idea of a HexChat‑style Copilot hit me so hard. It’s not just a clever concept. It’s the interface I’ve been missing. A multi‑channel, plugin‑friendly, terminal‑native AI client would give me the structure I’ve always needed: separate rooms for separate parts of my mind. A writing room that remembers my voice. A research room that remembers my sources. A daily‑log room that remembers my rituals. A project room that remembers my frameworks. Each channel with its own memory hooks, its own continuity, its own purpose. And all of it living inside the CLI, where my brain already knows how to navigate. It’s the difference between “AI as a chatbot” and “AI as tmux for my cognition.”
The terminal has always been the most cognitively ergonomic environment for me. It’s quiet. It’s predictable. It doesn’t freeze. It doesn’t ambush me with motion or noise. It gives me a stable surface to think on. When I’m in Bash or PowerShell, I’m not fighting the interface. I’m not being asked to split my attention. I’m not being visually overstimulated. I’m just typing, reading, thinking, and moving at my own pace. It’s the one place left where nothing tries to autoplay. A Copilot that lives there — in the same space where I already write scripts, manage files, and shape my environment — would feel like a natural extension of my mind rather than another app I have to babysit. It would be the opposite of the modern web, where half the CPU is spent fighting whatever JavaScript framework is trying to reinvent the scroll bar.
And the plugin idea is what makes it powerful. I can already imagine how it would feel to work this way. I’m writing something and want to open it in LibreOffice. I’m drafting notes and want to send them to VS Code. I’m working on an image concept and want to hand it off to GIMP. Instead of bouncing between apps, I’m in one quiet terminal window, and the AI is the connective tissue between all the tools I use. It becomes a cognitive command center instead of a chatbot. Not a productivity gimmick, but a thinking environment. A place where my executive function isn’t constantly being taxed by context switching. It’s the spiritual successor to the Unix philosophy: do one thing well, and let the pipes do the rest.
And the best part is that nothing about this violates how Copilot is meant to be used. It could absolutely exist as a third‑party client on GitHub. It wouldn’t impersonate Microsoft. It wouldn’t break any rules. It would simply be a different interface — one built for people who think in text, who need structure, who need calm, who need continuity. PowerShell on Windows, Bash on Linux, zsh on macOS. The same interface everywhere. The same quiet. The same clarity. The same sense of being in control of my own cognitive environment. It would be the first AI client that feels like it belongs next to grep, not next to TikTok.
This matters to me because the future of AI shouldn’t be louder, flashier, or more overwhelming. It shouldn’t be another sensory arms race. It should be more thoughtful. More structured. More accessible. More aligned with the way real human minds — especially neurodivergent minds — actually work. A HexChat‑style Copilot is the first interface concept I’ve seen that treats AI as a cognitive partner instead of a novelty. It gives me rooms for my thoughts. It gives me memory. It gives me continuity. It gives me calm. It gives me back the internet I grew up with — the one that made sense, the one that didn’t require a GPU just to load a news site.
I’m not imagining a toy or a gimmick. I’m imagining a missing piece of the computing ecosystem, one that fits perfectly at the intersection of neurodivergent cognition, early‑internet ergonomics, and the emerging role of AI as scaffolding for real thinking. This isn’t just a good idea. It feels necessary. And I’m exactly the person to articulate why.
Scored by Copilot, Conducted by Leslie Lanagan
Stories That Are All True 