$ man how-to/build-content-engineering-system

The win is not any single tool. It is building a system you fully control that gives you complete refinement over your content engineering pipeline at hyper speed. A CMS gives you a form and a publish button. A content engineering system gives you a codebase where every content type is a TypeScript interface, every page is a template, every link is programmatic, and every deploy updates three sites simultaneously. You change one data file and the knowledge base, RSS feeds, sitemaps, and schema markup all update in one push. No manual steps. No platform limitations. No vendor lock-in. Your knowledge graph is an asset you own, stored in version-controlled TypeScript files, not trapped in a SaaS database you cannot export from.

PATTERN

ShawnOS is a Turborepo monorepo with three Next.js apps (shawnos.ai, thegtmos.ai, thecontentos.ai) and a shared package. The shared package holds everything that crosses site boundaries: components, data files, RSS feed infrastructure, type definitions, and utility functions. Each app imports from the shared package. One component library, one data layer, one set of types. This means a new knowledge term added to engineering-terms.ts is immediately available on all three sites. A schema component update propagates everywhere. An RSS feed enhancement works across all feeds. The monorepo eliminates duplication and ensures consistency. You do not need three sites to benefit from this pattern. Even a single site gains from separating data, components, and app configuration into distinct packages.

CODE

Every piece of content in ShawnOS is a TypeScript object in a data file. Knowledge terms are objects with name, definition, whyItMatters, howYouUseIt, and related fields. Wiki entries have id, title, subtitle, category, description, keywords, sections, and related fields. The TypeScript interfaces enforce structure at build time. You cannot publish a knowledge term without a definition. You cannot create a wiki entry without sections. This is the content graph. The data files are the nodes. The related arrays and cross-references are the edges. Template-driven pages read the data and render HTML. You never write HTML directly. You write data and the system renders it. Adding a new content type means defining a new interface and a new template. The system scales horizontally without complexity growth.

PATTERN

The RSS infrastructure lives in packages/shared/lib/rss/. It uses the feed npm package to generate RSS 2.0, Atom 1.0, and JSON Feed 1.1 from the same data objects that render pages. Content source functions convert each data type (blog posts, wiki entries, knowledge terms, daily logs) into standardized FeedItem objects. A merge utility deduplicates and sorts items by date. shawnos.ai serves seven feeds: blog, all content, knowledge terms, how-to guides, daily logs, Nio terminal logs, and updates. Each feed is a Next.js route handler that calls getFeedConfig, converts the relevant data to feed items, and returns a Response with correct Content-Type and Cache-Control headers. Adding a new feed takes five lines of code. The entire feed infrastructure regenerates on every deploy because the data files are the source of truth.

PRO TIP

The end state is a system where adding content is adding data, not navigating a CMS. I open engineering-terms.ts, add a new term object, save the file, and push. On deploy, that term appears on the knowledge page, gets its own anchor link, appears in the RSS feed, gets schema markup, shows up in the sitemap, and is available for programmatic internal linking from every page that mentions it. Total time: under two minutes. This is the competitive advantage tools like AirOps complement but cannot replace. AirOps is excellent for competitive intelligence, citation monitoring, and share-of-voice tracking. Use it for visibility into how your content performs across AI engines. But the system itself — the content graph, the templates, the automation, the feeds, the schema pipeline — that is yours. Own the system. Use tools to monitor it. The system is the moat.