What is an LLM endpoint and do I need one?
An LLM endpoint is a public file that summarizes a website for software to inspect. It can aid discovery and checking, but it cannot guarantee crawling, use, or citation.
An LLM endpoint is a public JSON or text file that summarizes a website's identity, content, and related resources. It gives crawlers and other software a simpler document to inspect, but the visible website remains the main source people use.
The name is informal, and different systems use different files. This portfolio publishes generated summaries at /llm.json and /ai/llm.json. It also publishes /llm.txt, a manifest at /ai/manifest.json, and a Dataset catalog at /ai/catalog.json.
Do you need one?
You may benefit from one if your site has enough public content to summarize and you can keep the file accurate. A small site with weak or unfinished pages does not become useful because it added an endpoint. Build the useful content first.
Check three things before publishing
- It matches the website. The name, description, topics, and links must agree with what a person can see.
- It stays public and current. A stale summary can be worse than no summary because it sends software to old or missing information.
- It comes from the same source. Generate the summary from the site's current manifest and catalog instead of maintaining a second set of claims by hand.
An endpoint is evidence that the owner published a clear summary. Server logs may later show that a crawler retrieved it. Neither fact proves that an outside AI system used or cited it. The guide to knowledge assets and AI discovery explains those proof levels.