Hard world ware

Isolated environment for running LLM tasks locally and in the cloud.


You live in a world dominated by software. You consume a vast amount of time inside a program, both in the hard spending of tangible time, but also subconsciously your mind is constantly driven into thoughts about some remote database item (or so called tweet, or instagram post, or YouTube vid). What you experience is profoundly shaped by your online behaviour.

Lets look back at the world before the transition to the internet happened. A thought experiement might occur in the streets of New York. Lets imagine it being black and white. What you might see is a system of humans doing things. Walking, talking, all which the mind can imagine. What you will not see is compulsive staring at a titanium glass device with flickering lights. You might see people stand taller, without a curved neck.

Lets now do the same thought experiment, but now transition the black and white image you see into color. Make it feel like today. What does it look like? You will see the same remarkable system of people doing things. Walking, talking, and all which the mind can create of images. Now this time, you might see a few, or even many, people completely shut of from the world, either via staring onto a titanium device, or blocked off inside a pair of plastic headcups. Perhaps a few curved necks, bending to the will of the gravity, staring at the titanium device and its flickering delight.

But lets not get distopian just yet. Because, looking at this too lightely is not sufficient. With any technology there is a resulting positive, and sometimes a negative. And when the scale becomes so great (where anyone has a smartphone) the probability of negatives increases as well.

The negatives are largely driven by software. Not neccessarily the hardware. Its not what it is that matters, but what it does. Getting into the vast negatives will consume all of our time, but we must look at three core problems.

a) Our physiology is not designed to sit down for long hours and do work on a computer. We are simply not designed for it. One problem I have with software is its poor optimization for human physiology. This might be a spreadsheet app being designed for manual human labor, not for computer labor.

b) Our brain chemistry is getting tampered with to great extents by uncontrollable surges and crashes of dopamine sourced from the various entertaining information readily and cheaply available on any device, at any time.

c) Software must be considered art, any software written that does not invoke a sublime feeling in its human user must be deleted and rewritten.

My goal is to lay out these ideas to provoke a few considerations in your mind. Too often we are driven entirely by the conventional mind (mind saturated with the status quo, and remains outside the roam of original creative thinking), seeking not what we believe to be true, but what is comfortable to believe is true.

Designing software from first principles to work for, not against, our physiology is based on the idea that work should be orchestrated by the user, then delegated to preferrably the cloud, so the user can completely get of the computer and return to the present. This is contrary to traditional software that might gauge success of a product and user retention based of time spent in app, ie, active time. Usually tracked by minutes. What I believe is the superior metric to track is: what is the absolute minimum time per user spent, while still getting successful result with X?

Getting clear on what software success metrics matter to you is absolutely about original, first principles thinking. You have to avoid conventional metrics at all cost. Such measures are not driven by what you ought to believe, but by what you are programmed to believe and reinforced by comfort.


I am terribly scared of the effects on the group humanities mind. Throughout history you can clearly see the record of group think. What group think does is terrifying in retrospect. Yet, in the moment the inverse feels scary. Believeing something other than the group goes against your very human nature of adherence to a group, to secure your safety. Know this, study history, and observe simply the effects of blind faith in the groups mind. Through this studying read about the rise of the nazi party, and their ultimate faith.

The purpose of surfacing some history, is to provoke an idea in your mind. See, what you read, who you follow, and what you ultimately end up thinking is the cache of the information you have consumed, ideas that got iterated. Over time, being a naïveté, keeping blind faith of such external information manifests in you believeing whatever the information is to be true. Let this loop run for sufficient amount of time, and you become "programmed".

Now, step back for a moment, and think about your perceived identity. Apply some meta cognition to your thinking right now. What is something you believe to be true, deeply, that might keep you away from seeing other truths? Fundementally, this is not about looking for truths, but rather seeing everything there is from more perspectives or potential combinations than your immediate experience.

Now, connecting this idea with software. There are continously arising new technologies you can use. The combinations and configurations are virtually limitless. But because of mental model you now have of how the group mind works, naturally the group will cluster around certain ideas and ways of combining and configuring the new technologies. Because of your outside view, the observational mind, you can see this clearly. People cluster around the ideas, and you simply aknowledge this as data points in your head. From here, if working on applying this new technology, or a new idea, you must reason from first principles about the limitless potential combinations, and explore rigorously the various potentials.

This is a fundemental idea all of our software is built upon. We might use an "AI agent", but if you are a user of our app, you would never know. Instead, you would perhaps think about the increased speed, or the surprising abilities the tool has to do a task. See this, nobody cares about what your tool is, or what special souce you have. People do care about the things that it can do for them. How does it satisfy their desires. People has this desire, and you fail to provide clarity on what you can do to satisfy them or their desire, then you have utterly failed.

Never underestimate the degree to which people crave clarity and respond positively to it. - Ken Segall


Making software work for peoples minds is built up from how it works visually for users. Apps built for the entertaining effect are chaotic and flashing. Software made for deep work is built with a vastness of whitespace, yet the speed of a bullet. With any software, it is a highly satisfying feeling to add more things. To make it have more stuff. But behind this self indulgence lies a lack of empathy for the users who will spend time inside of the tool you have created. Fundementally as a creator, your task is to make art. While art is a creation for its own sake, you must optimze the art to not intefere with the easily influenced chemistry of the human mind. In essence, we want our software to live at the intersection between biology and art. How can we build something that does not deteriment the biology of a user, while smoothly delivering its promise?

Keys to the next generation of software

As you progress more and more into the chapter of history where computers can do the work of most tasks, keep the thought in mind about how you blend your first principles thinking with the technology at hand, to make something that allows for work to be done, with complete alignment with the human user biology. We are easily entertained, dont fall into the trap of making your tool do more entertainment that work. Let it be art, simply, while keeping its rigor its main chance at survival. All of the most valuable and brilliant ideas comes not from the group mind, it comes from the first principles thinker, who apply self reliance and faith in their ability to make things work. Let these ideas not become new truths, instead, let this become new data points in the vast space of brain power you inhabit.