*codeplain raises $3M around a bet on Phoenix Architecture
Today, we’re excited to announce that *codeplain has raised a $3 million seed round, led by GapMinder with participation from Silicon Gardens, to build a new software development workflow for the AI era.
Over the last two years, AI has fundamentally changed how software is built.
Coding agents are becoming dramatically better at producing software. Every new generation of models writes more code, solves more complex problems, and automates more of the development process. For the first time in the history of our industry, software is no longer constrained by the cost of writing code.
Instead, it will soon be constrained by the cost of maintaining it.
“Agentic coding is giving the industry 10x more code,” said *codeplain founder and CEO Dusan Omercevic. “But unless we fundamentally rethink maintenance, software teams are going to drown in the complexity created by that explosion.”
As we thought about this problem, we found ourselves increasingly aligned with the ideas behind Phoenix Architecture.
Coined by Chad Fowler, Phoenix Architecture offers a new mental model for software engineering in the AI era. It argues that as implementation becomes inexpensive to generate, source code should no longer be the primary artifact that developers and AI maintain. Instead, the durable layer of software moves upward into specifications and system intent, while implementation becomes something that can be regenerated whenever requirements change.
We believe that shift is inevitable.
And we’re building *codeplain to bring that model into production software development.
Our platform enables coding agents to work at the specification layer instead of the implementation layer. Rather than writing and maintaining source code directly, coding agents write and evolve specifications that describe system behavior and intent. *codeplain then transforms those specifications into tested, validated, production-ready software.
The result is a software development workflow centered on specifications.
Specifications are faster and more efficient for AI to produce than implementation code. They’re easier for developers to review, understand, and maintain. And because implementation is generated from specifications, software can continuously evolve without making the source code itself a long-lived artifact.
As Chad Fowler puts it:
“The industry is still thinking about AI as a way to write code faster. But faster code generation without a new maintenance model simply accelerates software entropy. Phoenix Architecture shifts the durable layer upward from code into specifications and system intent.”
The idea of treating specifications as the primary software artifact has deep roots in software engineering.
One of the most influential recent explorations was SpecLang, a research project from GitHub Next that proposed specifications, rather than source code, as the primary source of truth, with AI-powered tooling responsible for generating and maintaining implementation. Today, SpecLang creator Johan Rosenkilde is an advisor and angel investor in *codeplain, helping us bring that vision into production software development.
As Johan says:
“SpecLang was an early exploration of what happens when specifications become the source of truth instead of code. What *codeplain is building is one of the first serious attempts to bring that model into production software development.”
At *codeplain, we’re betting that software engineering is entering a new chapter.
The first wave of AI development focused on generating code.
The next wave will focus on generating and maintaining specifications, while treating implementation as something that can be continuously regenerated, tested, and validated.
The seed funding will enable us to continue extending coding agents with agentic capabilities for creating and maintaining high-quality specifications, while making *codeplain a faster, cheaper, more reliable, and more secure platform for transforming those specifications into production software.
We’re grateful to our investors for believing in this vision, and we’re especially excited to welcome Cosmin Ochisor, Partner at GapMinder, to *codeplain’s Board of Directors. We’re also grateful to our customers and design partners for helping shape the platform, and to everyone who believes software engineering needs a fundamentally new workflow for the age of AI. Above all, we’re grateful to the incredible *codeplain team whose talent, ambition, and relentless execution have made this milestone possible.
We’re just getting started.