Understanding ADP (AI Discovery Protocol)
The AI Discovery Protocol (ADP) is an open specification that Pressonify published to make websites systematically discoverable by AI systems. Think of it as a comprehensive sitemap for AI, not just listing pages, but explaining content relationships, update frequencies, and semantic meaning.
The ADP Endpoint Stack
A complete ADP v3.0 implementation includes endpoints such as:
- /.well-known/ai.json: Main manifest with capabilities and endpoints
- /llms.txt: Compact site context (under 2KB)
- /llms-full.txt: Extended context with content summaries
- /ai-sitemap.xml: AI-specific sitemap with semantic metadata
- /feed.json: JSON Feed v1.1 for content updates
- /updates.json: Delta feed for incremental crawling
- /knowledge-graph.json: Schema.org entity relationships
Why ADP Matters
Without ADP, AI crawlers must guess what your site is about by parsing HTML. With ADP, you explicitly tell them:
- What topics you're authoritative on
- How your content is organized
- When content was updated
- How pages relate to each other
ADP Compliance Levels
Sites can implement ADP progressively:
- Basic (40%): robots.txt + llms.txt
- Standard (70%): + ai.json + sitemap-ai.xml
- Complete (100%): All endpoints with HTTP headers
Key Takeaways
- ADP is an open specification built from machine-readable discovery endpoints
- llms.txt and ai.json are the core components
- ADP makes AI discovery explicit, not implicit
- Pressonify.ai runs the reference implementation of ADP v3.0
- ADP builds on crawlable pages and sitemaps with a structured discovery layer