The trap of building a software engineer

Building AGI means building something that can improve itself.

Since AGI is artificial, this literally means building software that can build software.

Which is why code generation is profoundly important for us to get to AGI.

At the same time, the problem when you pursue code gen is that you start to go too meta.

Yous start building code gen for your code generation software.

And you lose grasp of why you're building the software engineer in the first place.

I learn my lesson. I incorporate "basic software" into my day to day.

That's why I think the best way to build a software engineer is to build business applications: this keeps you focus on the right things.

And this is why the best software engineer cannot be sold as a software engineer first. You have to pursue something more business oriented.

The trap of building code generation tools is to measure your success by how many people use your tool.

The real indicator is whether you can build business appplications with it. Because if you can, you have the potential to completely disrupt the B2B SaaS market.

But it doesn't end here: you need to build software that drives itself. Something that makes the work flow.

And once you have this, you need to incorporate a growth loop into it. Because your software won't sell itself.

The explosive combo is:

1- a product that has software engineering capabilities. 2- the software produced can be consumed very easily by end users because it knows the next step. 3- the software can find who needs it and reach out to these people.

The ultimate AGI is a self-running business. We're all after this: building a freelancer.

Can you imagine what happens once a digital freelancer does the work of 10 people AND knows how to find new customers?

It will eat the lunch of 1000s of companies. Fat will be trimmed. Innovation will accelerate. And we'll start solving more important issues than software engineering.

Software was always a means to an end. A bigger problem is energy. I really hope that "solving software" will help us innovate faster in the "real industries".

Let's all commoditize software.