Sending articles to the AI knowledge base

Your help centre and your AI agent's knowledge base are two different things. Publishing an article does not put it in the agent's index, and nothing puts it there automatically. Handing an article to the AI is a deliberate act, and this article is about that act.

There are two controls: an include/exclude flag on each article, and a bulk send from the Content list.

Two things have to be true first

A connected agent. The articles go to this workspace's Aura AI agent, and if there is none connected there is no knowledge base to send to. Connect one under Aura AI. You cannot point the send somewhere else from here; changing the destination means changing the connection.

A public address for the help centre. This is the constraint that surprises people, so it is worth being blunt about the mechanism:

The indexer fetches your articles over the public web. It is given the article's URL, not its text. It then visits that URL like any other reader.

Two consequences follow. A help centre with no public address yet has no URLs to hand over, and nothing can be sent until you set one under Knowledge → Domain. And an article that is hidden or internal cannot be indexed at all — the indexer arrives signed out, gets a 404, and indexes nothing. Auralis refuses those before sending rather than letting them fail later: "Not public. The indexer fetches the article over the public web, so a signed-in or internal article would index as a 404."

The same is true of drafts. Only a published, public article can go to the AI.

The include/exclude flag

Each article's settings rail in the editor carries a toggle for the AI knowledge base.

It is an exclusion flag, not a send button. Turning it on does not hand the article over; it puts the article in the set that a later, explicit send is allowed to include. Turning it off marks the article as not belonging in the index.

And excluding never deletes anything from the AI. If an article is already live in the index when you exclude it, Auralis flags it as needing removal — the chip reads Remove from AI — and leaves the actual removal to you, in the AI service. Nothing disappears behind your back.

Sending from the Content list

Knowledge → Content shows every article with its AI state. Tick the ones you want and a bar appears reading Send N to AI.

There is deliberately no "send everything" button. What is selected is what is sent. Your selection follows you across pages, so you can filter, page and tick without losing it.

Confirming shows you the agent it is going to, by its full identifier, and repeats the rule: only public, published articles are sent, and anything in the selection that cannot be sent comes back listed with its reason, having never left.

The report afterwards ends with the sentence that matters most: sent is not the same as searchable. The AI service confirms indexing separately, and the AI column updates when it does.

The seven states

• Not indexed — never handed over.
• Sending — handed over, and the AI service has not been asked about it yet.
• Pending — the AI service accepted it but has not confirmed it is indexed. Accepted is not the same as retrievable.
• Indexing — the service reports it is still working on it.
• Indexed — the service confirms it holds the article and can answer from it.
• Send failed — the last attempt did not work. The service's own words are shown with the article.
• Out of date — the version handed over is not the version published now.

Out of date is the one to build a habit around. It appears whenever you edit and republish an article that is already in the index, because editing an article does not re-send it. Nothing else re-sends it either — no import, no publish, no visibility change. Sending again is always something a person does, and the state exists so you can find the articles that need it.

If confirmation never arrives, Auralis eventually stops waiting and marks the article failed. Read that as "no confirmation came back", not as "the content is missing".

A workable routine

Send in one pass after your first publish. Then, once a week or once a sprint, filter Content by Out of date, tick everything, and send. That is the whole maintenance story, and it takes a minute.