Building a native publication layer

Publications now have a first-class place inside ComputeIntel. The immediate goal is simple: give founder-authored thinking and institutional work a shared home without flattening them into a generic company blog.

The first version is intentionally repo-authored. It uses the same static delivery model as the rest of the site, keeps the content close to the product and avoids introducing a CMS before the publishing workflow requires one.

The publication contract

type Publication = {
  title: string;
  slug: string;
  author: AuthorHandle;
  contentType: PublicationType;
  topics: string[];
  status: "draft" | "published";
};

Authors are referenced through a registry rather than embedded into every record. That keeps attribution consistent today and leaves a clean path for a selected group of contributors later. A publication type is metadata, not a separate page system, so an essay and a build log share the same reading experience.

What the first version does not do

• No public contributor sign-up or self-service publishing.
• No database, browser editor or external content platform.
• No author archives, topic routes, comments or membership layer.
• No dependency added solely to transform article content.

These are deliberate boundaries, not missing ambitions. The content model can support future archives, feeds and research artefacts, while the first release remains easy to understand and operate.

The wider Applied Intelligence Protocol remains the organising system. Publications are where its ideas, evidence and updates can be developed in public.