Articles are the primary Q&A content type in AI Cloud. Each article is associated with one or more intents and contains answers that are returned when a user's message matches those intents. Articles support multiple answers with context conditions, allowing the same question to return different responses depending on the state of the conversation.
Structure
-
Intents — the questions or phrases that trigger this article, managed in the Recognition section.
-
Default answer — the answer returned when no context condition matches, or when no contextual answers exist.
-
Contextual answers — additional answers, each tied to one or more context variable values. When the context matches, the contextual answer fires instead of the default.
-
Labels — optional tags for organisation and filtering.
Answer Types
Each answer (default or contextual) can be one of several output types:
-
Text / channel output — formatted content delivered to the user via the output editor. Supports rich content depending on the channel.
-
Add dialog — routes the conversation into a CAIC Dialog flow.
-
Add agent — hands the conversation to a HALO agent. The agent handles subsequent turns until it completes, then control returns to the NLP flow.
-
Add tool — invokes a HALO tool with mapped input parameters. Output can be stored in a conversation variable. Control returns to the NLP flow immediately after the tool completes.
→ Using HALO Agents and Tools in Content
Context Variables and Metadata
Context conditions and metadata can be applied per answer using the Context variables / Metadata button. This controls which answer fires based on what context the client passes in with the request.