Apparently I've Been Doing Technology Ass Backwards
I've been screwing around lately with a Mudra Link, some browser code, a local WebSocket, and a little project I called the "Neuro/Bio Interface Lab".
That description probably makes this sound much more sophisticated than it is. It isn't.
This is a hobby experiment.
I'm basically poking at a weird wrist computer, looking at signals, drawing graphs, making noises, and asking increasingly unreasonable questions about what else it might be able to do.
Which is pretty much how I end up in most of these rabbit holes.
But somewhere in the middle of this one, I realized there may be a larger pattern to how I've always approached technology. I think I do it ass backwards. Most people get a new piece of technology and ask: what can this do? My first reaction is usually closer to: why the hell can't it do this other thing? And then I get annoyed.
This isn't necessarily a personality trait I'm proud of.
My relationship with technology may largely be based on selfish arrogance and unrelenting demands.
I don't want to learn how the machine wants me to work. I want the machine to figure out how I work. If it doesn't, I start trying to bend it until it does.
That probably explains a lot.
The Interface Thing
Human-computer interface is one of those subjects I've always found fascinating because, ultimately, it's the whole ballgame.
It's how we actually use all this stuff. The processor can be incredible. The model can be incredible. The software can be incredible. But at some point a human being still needs to communicate: I want something to happen.
For most of computing history, we've built layers that translate that desire into machine-readable instructions.
Keyboard.
Mouse.
Touchscreen.
Menus.
Icons.
Command lines.
Apps.
Then AI gave us language. That felt like a huge change because suddenly we could just ask. Except we immediately invented prompt engineering.
Which was basically: now you can talk to the computer naturally, as long as you learn the unnatural way to talk to the computer naturally.
Then came context engineering.
Instead of endlessly trying to phrase everything perfectly, we started giving the system more information about what was going on.
History.
Files.
Memory.
Tools.
Environment.
The current task.
What happened five minutes ago.
What normally happens next.
And that made me wonder if there's another layer forming now. Maybe something like intent engineering. I don't know if that's the right name.
I don't know if anyone needs another "engineering" term. We may have enough already. But the idea is interesting.
The Prompt Was Never the Point
The prompt isn't actually what I want. It's a description of what I want. A mouse click isn't what I want either. Neither is a swipe. Neither is a gesture.
Those are all just ways of expressing intent to a machine that otherwise has no idea what I'm trying to accomplish.
That's what made the Mudra experiment interesting to me. The obvious use for something like this is gesture control. Move your hand. Pinch something. Rotate your wrist. Computer does thing.
Cool. But almost immediately I started thinking: okay, but what else is in there?
The device exposes motion, pressure, inertial data, gestures, and surface neural-channel signals. And I have absolutely no idea what the useful combinations of those things ultimately are. That's the fun part.
Maybe a gesture is enough. Maybe it isn't. Maybe motion plus pressure plus context gives the system a much better clue. Maybe the application I'm using matters. Maybe what I did two seconds ago matters. Maybe several weak signals together say more than one explicit command.
At that point, the input stops being: do X. It starts becoming evidence that: the human probably intends X. That's a much more interesting problem. And to be very clear, I'm not claiming I've solved any of it.
I'm barely claiming I know what I'm looking at. I have graphs.
That's roughly the current level of scientific achievement.
Apparently I Keep Trying to Remove the Interface
A few days ago I wrote that apparently I've been trying to remove the interface. The more I think about that, the more it fits.
A lot of things I've been playing with over the last few years looked unrelated.
Voice.
Agents.
Persistent context.
Local AI.
Computer vision.
Wearables.
Gesture interfaces.
Ambient systems.
But they all seem to point toward the same basic desire: reduce the distance between "I want something" and "it happened." Every interface layer in between exists because the machine needs help understanding us.
Open this... Click that... Type this... Choose that menu... Write the prompt... Select the tool... Tell the agent what application to use.
It's all translation.
And apparently my reaction to translation overhead is just increasing levels of irritation. I don't want to manage the machinery. I want the machinery to understand the outcome. That sounds obvious when you say it that way. 0The hard part is everything underneath it.
Maybe the Ultimate Interface Is Just Wanting Something
This is where the whole thing gets slightly absurd.
Traditional computing: human, then interface, then software, then action.
AI: human, then language, then intelligence, then software, then action.
Context-aware AI starts moving toward: human, then intention plus context, then intelligence, then action. And then you start wondering how much of the middle can disappear. Maybe someday the interface is mostly: want something.
The system figures out the rest.
Which model.
Which tool.
Which software.
Which device.
Whether it needs more information.
Whether it should ask first.
Whether it already knows enough.
That's obviously a very long way from a little JavaScript page drawing Mudra signals on my screen. But this is how I like to explore this stuff.
You start with something real.
You poke it. You misuse it. You ask dumb questions. Occasionally one of the dumb questions turns out not to be dumb.
So Anyway, Here's the Repo
I put the Neuro/Bio Interface Lab on GitHub.
It connects to the local Mudra Studio stream and lets you poke around navigation, gestures, pressure, direction, IMU, SNC, spatial visualization, raw packets, and some experimental audio feedback.
It's absolutely not a medical tool. It's not a scientific interpretation layer. It's not me claiming I know what nervous systems are doing. It's me saying: this thing is exposing interesting signals, let's screw around with them.
If you have a Mudra, build something.
Fork it. Break it. Add recording. Train a classifier. Turn it into an instrument. Make a game. Map it to an agent. Do something ridiculous I haven't thought of.
Have fun.
And seriously, if you build something interesting with it, let me know.
Because I have no idea where this shit goes.
Which is exactly why I'm interested.
Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.
Comments