Every product is either Cursor or Claude Code.

All productivity tools exist on a Cursor <-> Claude Code spectrum.

3 years ago, I wrote about the Neuralink <-> GPT Coordinate Plane.

In short, I claimed that all productivity tools (even pre-computers) can be mapped on a plane with Neuralink and GPT as the axes. I already try to maximally compress my thoughts so you should just read the post.

As technology has evolved, another dimension that has emerged. Rather than give this dimension a complicated jargony name, I will refer to the directionality of the axis by two tools that do the same thing in very different ways: Cursor and Claude Code.

To quickly define what those two tools are right now (and future proof this post):

Cursor is a code editor that lets you use AI to help write code. First, Cursor was mainly used for its tab auto-complete, allowing you to write code faster. Now, there is an AI that will take your prompt and edit files¹. Most importantly, the most prominent real-estate in Cursor is the code. You are staring at the code (the underlying thing you are manipulating) as you use the software.

Claude Code is different. Claude Code is a chatbot. When you start the product, you see nothing other than an input box waiting for you to type. Although Claude Code sits on top of a massive repo, you don't see it. When Claude is working, you mainly see its thinking traces. You don't interact with the code at all. You CANNOT interact with the code in Claude Code.

The end output is the same: you generate code (hopefully it's good). The methods of achieving this outcome are extremely different. I wrote this post in Cursor. It would not make sense to write this post in Claude Code - unless I didn't want to type a single character of the actual content (I don't frown upon AI writing like most, as long as it's good. But this article I wanted to write by hand.). But I used Cursor to fix typos and create footnotes.

My claim is that this dichotomy is exemplified by the status quo of productivity tooling. In every tool, a decision has to be made: will the user interact with the material? Or will they interact with the thing interacting with the material. "And so on."²

For the most part, the Cursor end of the spectrum represents the old guard of tooling: Notion shows you the document, and the Notion agent operates on that document.

So what is my big claim? A blog post has to have a big claim, right?

My claim is that the Cursor <-> Claude Code is equivalent to the spectrum of consciousness.

Whoa! 0-100 real quick... I was with you when you were talking about coding tools, Sam, but this is too far.

To close the loop, and maybe land the plane (you be the judge):

As productivity tooling moves along the Cursor <-> Claude Code spectrum, it also moves along the consciousness spectrum. An ant is closer to Cursor than a human is. An ant is closer to the bare metal³.

Consciousness is not an objectively better thing. Vim was a better code editor than VSCode for many people (if you were one of those people I want to hire you probably).

But consider that there is an incessant tide that pushes us as a species to develop more and more conscious tools. I am playing our part in building The Sentience Company, but even Cursor is becoming more Claude Code and less Cursor…⁴

Sentience is a Claude Code product. Sentience does not expose the guts of the system to the user. There is no direct manipulation of data. You are not moving bits or electrons around, nor code, nor markdown files. We are opinionated about what we show to the human, and that is often different than what the agent itself is manipulating or seeing. Often the human interacting with a Sentience sees nothing other than the thing they need (textual or visual — working on more dynamic outputs).

This is like interfacing with a conscious thing. The conscious thing actually doesn't experience the raw underpinnings. We cannot sense our neurons firing. But we feel the composite of these things, and can form memories and knowledge and share ideas from these composites. We have awareness (in Zen: Single Awareness).

Claude Code is great, but it makes us atrophy the skill of writing code. Perhaps the same can be said about us as humans — we've lost the art of metabolizing oxygen, instead merely outsourcing it to our cells! Bacteria had it right! They knew the true art of doing it themselves, instead of us humans who merely delegate our cellular agents to think for us.

My advice: ignore the bacteria-pilled haters. Build consciousness. And if you want to build consciousness Sentience with a team in NYC reach out to me sam@sentience.com.