The Recipe for Solving Software

Recursive graph based agents with a memory system are the recipe for solving whatever you are trying to build within software.

Anything else is distraction.

It reminds me of how everyone has been building companies in software while discarding the basis of building in the first place.

We have a black swan event around us. When we become too objective about what has happened, or about how software has worked until now, we will find ourselves in places where what we have built is entirely wrong.

We will come back to it. Some will be fortunate enough to survive this and get another shot at building, taking it all. Then there will be everyone else, destined to die sooner from the moment of inception.

This is the best time to solve software. But you have to understand why software exists in the first place. Look around, and within, for its recipe. Only build for that.

Solve Entirely

Whatever you are solving, solve for it entirely.

Solve for markets. Solve for humans. Create new markets.

Solve big or go home.

The entire race for AI has been to build for this technology. But we have to build to serve humans. We have to enable them to do more, to think creatively and freely.

Software is an accelerator for ideas one has about the world. It lets more people come together, collaborate on those ideas, advance toward more efficient systems, and exchange value within them.

Why Software Became The Arbitrage

We had limited operational bandwidth.

Then information was digitized with computers. A bigger arbitrage than software and technology existed: humans.

These people came to be known as knowledge workers. One could trade time and skill for money. Even for the same set of skills, the people who could project their ownership better took a larger share of the value created from software.

A lot has changed since. With AI, accessibility has changed the most, and it is changing with unprecedented speed.

More has happened in software than in the entirety of software history, but ownership has stayed the same.

AI is simplifying your understanding. When execution is no longer the bottleneck, created and shared value flows through other limited resources.

Internet democratized knowledge. AI is democratizing intelligence.

Even with the same amount of resources, the only distinguishing factor apart from luck is your projection of understanding and empathy for what you are building, and how many people it empowers.

Back To First Principles

When building is solved, we go back and retrospect on what needs to be built.

We take a pause and understand why it all exists in the first place.

The technology that has served people has stayed. The technology that serves other technologies to be built upon has stayed relevant for longer.

So the question is not just what you can build now.

The question is how many people give a fuck about your company, and how much value is transacted within the systems you build.

That is the only thing that matters.

Originally posted on X.