Ask a question about Northmoor University and get an answer based on the site's own content, including courses, programmes, departments, events, facilities, news and staff profiles. The sources used for the answer appear alongside it.

Enter a question on the demo page, or choose one of the suggestions. The answer appears a few sentences at a time. Each claim has a number, such as [1], which corresponds to a source in the list below. You can also ask follow-up questions without repeating the earlier context.

How it works

The system turns your question into an embedding and compares it with a vector index of the site's content. It then gives the closest matching passages, along with your question, to an agent. The agent can cite only those passages, so it cannot fill gaps with general knowledge about universities. If the site has no relevant content, the system says it couldn't find anything rather than guessing.

Things to try

Start with a question the site clearly answers, then try one it does not cover and compare the results. You can also ask a follow-up that depends on your first question. If an answer seems wrong, check the Sources list. It should show whether the model misunderstood the source or the source itself was unclear.

Part of a larger toolkit

AI Answers doesn't work in isolation. It works with the site's overall Guardrails setup and with whatever context has been configured site-wide, the same way every other AI feature here does. It's also not tied to one vector database: this site indexes into Postgres, but the same feature works against any supported vector database provider.

Learn more

This demo uses the AI Answers module. For easy setup, we also have an AI Answers recipe. Full documentation is available at project.pages.drupalcode.org/ai_answers.