The email arrived on a Wednesday, from a researcher at a university Justin had heard of but never visited. The subject line was direct: "Request for Access to Software Factory Methodology." The body was polite, formal, and three paragraphs long. The researcher was studying AI-assisted software development practices. They'd read the factory.strongdm.ai page. They wanted access to the internal methodology, the process documentation, the tooling specifications. They were prepared to sign an NDA.
Justin's reply was four sentences.
He thanked the researcher for their interest. He explained that the factory's methodology was not proprietary. He sent a link to the Attractor repository on GitHub. And he wrote: "It's all there."
Jay saw the exchange because Justin forwarded it to the team channel with no commentary. Jay read the researcher's email, then Justin's reply, then clicked through to the Attractor repo.
The repo contained three specification files. attractor-spec.md. coding-agent-loop-spec.md. unified-llm-spec.md. Together they were 5,700 lines of pure specification. No code. No proprietary algorithms. No secret sauce hidden behind an API. Just markdown files describing, in precise natural language, how the factory's pipeline runner worked, how the coding agent loop operated, how the LLM integration was structured.
The spec was the product. The product was the spec. There was nothing behind the curtain because the curtain was transparent.
"The researcher wanted to sign an NDA," Jay said.
"I know," Justin replied.
"They thought the methodology was a secret."
"Most people think the methodology is a secret. They assume that if something works, the details must be proprietary. The opposite is true here. The details are public because transparency is part of how it works."
Jay thought about this. The NLSpec approach—natural language specifications directly usable by coding agents—relied on clarity. Specs that were written to be understood by both humans and AI agents had to be precise, unambiguous, and complete. Hiding them would defeat their purpose. The specs worked because they were clear enough to be public.
"There's no moat," Jay said. Not as criticism. As observation.
"The moat is execution," Justin said. "Anyone can read the specs. The advantage is that we've been running them for months. We have the scenarios, the satisfaction history, the institutional knowledge of what works and what doesn't. The specs are the starting line. We're miles past it."
The researcher replied the next day. They thanked Justin for the link. They said they'd expected a process involving legal review and access agreements. They hadn't expected a public GitHub repository.
Justin didn't reply to the follow-up. The link was the reply. It's all there.
"It's all there." Three words. No NDA, no access agreement, no gated documentation. Just a link to a public repo. That's radical transparency and it's the right play. The moat is execution, not information.