Preface: Towards Software Design

I want to make the SaaSpocalypse real. This website is dedicated to understanding the many obstacles in our path, plotting a course around them, and discovering a new way to build software.

When I say SaaSpocalpyse, here's what I mean. Start with the notion in the business press that existing software firms are bloated, slow, and deliver expensive solutions that never quite meet the needs of the target audience, despite ever increasing budget and headcount; these firms are at truly at risk from younger firms that thoughtfully integrate AI into their workflows. But beneath the macroeconomic layer, inside the individual software team, a revolution is coming, one that will overthrow technology for technology's sake and elevate humanistic product design in its place. Teams will support more product than ever before, either in breadth (more features) or in depth (richer ones).

None of this will happen automatically. We must define the vision, iterate it, and propagate it, all while pushing against decades of status quo thinking. There are two fronts:

1. The SDLC. Today's Agile ignores too many important decisions until coding, and it permits too many poor decisions that don't align with the product. Like an inefficient engine, vast amounts of energy are translated into waste heat instead of forward motion.
2. Coding. Every engineer's goal should be to never read or write a line of code again, in the same sense that they don't read or write assembly code. This tooling layer does not exist, and must be created.

Is this new world possible? There is good cause cause for skepticism, given how AI boosters make similarly extraordinary claims, with the exact same tone they used to promise web3 and the metaverse. Real-world evidence suggests AI coding is a disaster: security flaws, production incidents, and vibe coded projects that cannot grow out of the prototype stage.

But yes, this world is possible, and we already see it today: when agentic software production goes well, it is incredible. The challenge is to understand why it works in some scenarios but not others, and from there broaden the conditions for success to not just coding, but the design of product specifications. We start with fundamental questions, such as:

• What does a software team actually do?
• What is an LLM, and how does it help us?

Then, we can form mental models and techniques which:

• Flush out product requirements early and in a rational order,
• Thoroughly test product design and find flaws early,
• Create LLM-first tooling that enables the graceful evolution of prototypes into reliable, malleable production software.

The coming revolution in software production will be felt throughout the economy, including on the individual level. The software engineer will go extinct, and the software designer will take its place. Unlike software engineering, software design is humanistic -- the hardest part of the job isn't writing code but understanding what the user wants, meeting these desires with experiences that satisfy them. The resulting software will be richer, freer, more tailored to the communities it serves, and our society will be better off for it.