# SEO and answer engine optimization for the texts service

A round of changes to make the texts service more visible to both search crawlers and AI answer engines.

## What landed

**Open Graph tags** — every page entry now emits `og:title`, `og:description`, `og:type=article`, and `og:url`. This drives link previews in Slack, iMessage, and social platforms. The description is extracted from the first non-heading paragraph of the post.

**Canonical URL** — `<link rel="canonical">` on all HTML responses. Important because the same URL serves two formats (HTML and text/plain, negotiated via `Accept` header). Without a canonical hint, crawlers might treat them as duplicate content.

**Semantic HTML** — page entries are wrapped in `<article>` inside `<main>`. Dates use `<time datetime="...">` with ISO timestamps. Listing footers use `<nav>`. These are minor ranking signals but free to add.

**`robots.txt`** — previously 404'd for crawlers. Now serves a permissive file pointing to the sitemap.

**`sitemap.xml`** — all public page entries with `<lastmod>` dates. Helps crawlers discover content that isn't densely interlinked.

**JSON-LD Article schema** — each page entry includes a `<script type="application/ld+json">` block with `headline`, `datePublished`, and `url`. This is the primary AEO signal: Perplexity, Google AI Overviews, and similar systems use structured data to identify citable content and extract metadata without parsing prose.

## SEO vs AEO

Traditional SEO optimizes for PageRank-style signals: backlinks, keyword density, crawl budget. Answer engine optimization targets a different extraction pipeline — AI systems that summarize and cite sources directly. The signals overlap (structured data, clear headings, canonical URLs) but AEO adds a layer: content that reads as direct, declarative statements is more likely to be quoted verbatim.

The `llms.txt` file was already doing AEO work — it's a machine-readable index designed for LLM consumption. The JSON-LD and OG tags extend that to the web layer.