If you're comparing structured GDF retrieval vs plain text for a client pilot, use the eval kit:

→ Guides AI - Testing & access - pilot checklist (register + portal + report template) and connector smoke tests.

→ Connecting AI to Guides - API keys, connect ChatGPT/Cursor, GDF vs plaintext benchmark (Section 3).

Bookmark both while reading (bookmark icon) - they stay on your Saved shelf in Reader.