Skip to content

Generative Engine Optimisation (GEO)

This governance document establishes the standard for architecting all documentation within the DSOM framework to be natively "AI-ready." The digital landscape has shifted from traditional deterministic search (SEO) to probabilistic information synthesis (GEO/AEO).

1. The GEO Paradigm

Answer engines (e.g., ChatGPT, Perplexity, Claude, Google AI Overviews) synthesize contextualized answers directly. To ensure our documentation is extracted, cited, and summarized accurately, we must optimize for machine readability, source attribution, and direct answerability.

2. Empirically Validated GEO Strategies

All human contributors and AI agents must actively employ the following content strategies when drafting documentation: - Quotation Addition (+41% lift): Provide discrete, attributable human expertise. - Statistics Addition (+31-38% lift): Replace vague qualitative prose with specific, verifiable data points. - Cite Sources (+35% lift): Inject inline references to credible third parties. - Fluency Optimisation (+26% lift): Improve stylistic clarity to make text mechanically easier for the LLM to parse. - Authoritative Tone (+21% lift): Remove hedging and uncertain phrasing, which models interpret as low-confidence data.

Note: Keyword stuffing actively harms visibility by triggering AI trust-layer penalties (-8% to -12% drop).

3. Content Creation & Hierarchy for AI Ingestion

  • Atomic Intent: Documentation must consist of atomic pages with a singular, clear intent.
  • Context Window Chunking: Keep sections concise (max 200–400 words) to prevent the model from truncating vital information.
  • Semantic Hierarchy: Formulate H2 subheadings directly as common user questions (e.g., "How to rotate an API key").
  • Colocation of Concepts: Place examples (code snippets, inputs/outputs) in immediate proximity to the theoretical explanation to reduce cognitive load.

4. Machine Readability & The llms.txt Standard

LLMs consume raw text. Parsing HTML and complex DOM structures wastes computational resources. DSOM strictly enforces Markdown as the native language of knowledge representation. To guide AI crawlers, the repository must maintain an llms.txt specification at the root directory. This plain-text file acts as an XML sitemap for machine intelligence, providing a curated, high-signal map of our content.

Deep State of Mind (DSOM) For My AI Protocol | Harisfazillah Jamel (LinuxMalaysia) | 2026-07-11 Standard: UK English | DBP-standard Bahasa Melayu Malaysia (Piawai) | GNU General Public License v3.0