The Neuralink <-> GPT Coordinate Plane

All productivity tools exist on a Neuralink <-> GPT coordinate plane.

This framework best encapsulates the past, present, and future of productivity software and is particularly useful for those seeking to develop such products.

First some cursory definitions: Neuralink is a brain-computer interface technology aimed to reduce the barrier between human brain output and computer input, allowing for seamless interaction and communication between the two. Over the long term, Neuralink epitomizes the direction of technology that eliminates the input lag or friction from the human brain into a computer (or other augmented interfaces such as driving a car and operating a bionic arm). In otherwords, software on this end

GPT likely needs less of an introduction. In contrast to the Neuralink dimension, GPT epitomizes the “intelligent assistant” direction of software. Software on the GPT end of the spectrum augments the human brains ability to do work. To be clear, a case can still be made that GPT increases throughput (e.g. convert all of these rows from Celsius to Fareinheit), but the main breakthrough of these LLMs is their capacity for augmenting a single human brains generative output into that of many many human brains.

Let’s examine the current landscape of productivity software and see how they fit into this framework. In order to paint a clear picture of this spectrum, I want to outline a few examples of technology on each end of the spectrum, going all the way back to basic technology and bringing things forward towards the present/future:

The mouse: before the mouse, UI interaction was gated by how fast you could type (and memorize the hardcoded commands of what to type). As much as my Vim purist coworkers would like to claim keyboard supremacy, the mouse undoubtedly increases the rate at which the human brain can communicate with the computer.
Superhuman: the ethos of this email client is speed and efficiency. Everything is hotkeys. Everything happens fast. The inherent value-add is that it takes less finger movement to accomplish what you want.
Word, Powerpoint, Excel, Outlook, status-quo PDF Readers: The value of Word (pre-GPT integration) is simply that it is 1,000x better than a typewriter which in turn is 100x better than pen and paper. It speeds up the rate at which you can input your thoughts onto a page.
Figma: same thing. If you really wanted to, you could do everything Figma does on a crazy arts and crafts table.
VR/AR: I’ll focus here on the productivity aspect of VR/AR (the aspect which I’m most bullish on). VR allows you to consume information at a higher bandwidth than a rectangular screen. Via movement, hand-gestures, and facial expressions, VR also allows you to output more information per second than just a keyboard and mouse.

Now, let’s look at that which falls on the GPT side of the spectrum:

Books: trivial, yes, but the concept of a second brain has existed well before computers. While pen and paper don’t inherently generate new content, writing things down has always been humanities advantage over every other animal. Imagine if humans had to relearn everything over and over again each generation — we’d barely be able to start a fire. Instead, no single person needs to re-learn anything beyond some marginal improvement from one generation to the next. Reading a book from a person you’ve never heard of and will never meet is just like GPT, you’re outsourcing and leveraging knowledge. A library is like having a thousand subject matter expects at your beck and call!
Programming languages: compilers were the first “AI”. Write some text, get an output so far beyond what a human brain could do — that’s software!
The computer, the internet: the obvious augmentation of a pen and paper is a world where lot’s of virtual pens and paper can store information on a huge scale.
Notion: the natural extension of a second brain. Notion’s superpower is not that it is a Markdown editor. That is trivial. Instead, it is that it creates a natural parking lot for the human brain — or often, many human brains together.
GitHub: the second brain of code. Issues, PRs, READMEs, and Github Releases represent the knowledge base of an app.

The final question that underlies this post is: why frame things in this way at all? (or, as my 10th grade history teacher always said — “So what? Why is this important?”) The answer is that this framework has two very beautiful properties:

Current productivity software has emerged into these two categories naturally. I claim that applying a k-nearest-neighbor clustering on all productivity software would currently very neatly categorize into these two groups. This is emergent behavior and not something that people have intentionally pursued (obviously… sincewe didn’t even know what GPT and Neuralink were until a few years ago).
GPT and Neuralink as figureheads of each category represent two of the most exciting future looking technologies.

In other words, everything that we’ve created in the productivity space over the last 40 years has been a rudimentary version of either GPT and Neuralink this whole time — we just didn’t know it yet! The deeper point here is that these two axes represent the fundamental categories that humans strive for in their quest to improve productivity.