Vision statement

The Manifesto.

Abstract

For more than seventy years, software development has been organized around an assumption that was both necessary and extraordinarily successful: humans write instructions in languages humans can understand, and software tools translate those instructions into representations machines can execute.

Programming languages, compilers, interpreters, assemblers, intermediate representations, linkers, runtimes and development environments evolved around this relationship.

Artificial intelligence introduces a fundamental change. Increasingly, the entity creating software is itself a machine.

AI coding systems can already interpret human intent, reason about software architectures, generate source code, test applications, identify defects and modify programs. Yet despite this transformation, these systems continue to operate largely inside a software-development architecture designed for human programmers.

An AI system generates Python, C++, Rust, JavaScript or another human-readable programming language. That source is then parsed, compiled or interpreted, transformed through one or more intermediate representations, optimized, linked and eventually converted into instructions that a processor can execute.

The Irwin Machine Company is exploring a different question: If the programmer is a machine and the ultimate consumer of the program is also a machine, why must human-readable source code remain the language in between?

Irwin is investigating a machine-native software development architecture in which artificial intelligence and other autonomous systems may create executable software without conventional human-readable source code serving as the primary program representation.

The objective is not simply to create another programming language, compiler or AI coding assistant. The objective is to investigate a new computational development layer designed around machines writing machine code for machines.