Breadcrumbs

Conversational AI Cloud integration

AI Cloud serves as a proxy layer that orchestrates multiple NLP pipelines, including HALO. Context can flow bidirectionally between HALO and AI Cloud, enabling unified state management across both systems.

This integration allows you to:

  • Initialize HALO context from AI Cloud variables

  • Persist HALO context changes back to AI Cloud

  • Access HALO context values in AI Cloud API responses

How It Works

AI Cloud → HALO (Inbound Context)

When AI Cloud sends a request to HALO, it compares its conversation variables and context variables against HALO's context variable names.

AI Cloud Variable

HALO Context Variable

Result

customer_tier = "gold"

customer_tier exists

HALO uses "gold" as the value

  • If a matching context variable name exists in HALO, the value from AI Cloud is applied.

  • HALO stores this value and uses it throughout the current interaction.

  • The value persists for future interactions unless explicitly changed.

HALO → AI Cloud (Outbound Context)

When HALO returns a response, all context is passed back to AI Cloud.

  • AI Cloud maps returned context to conversation variables of the same name.

  • These values are assumed for all future interactions in that conversation.

Example:

  1. HALO sets context name = "Sarah" during an interaction.

  2. AI Cloud receives this context and updates its conversation variable name to "Sarah".

  3. On subsequent interactions, AI Cloud will send name = "Sarah" to HALO automatically.

Context in API Response Metadata

When AI Cloud processes a response from HALO, it includes all HALO context variables in the metadata object of its API response.

Note: Context is only included in metadata when AI Cloud uses HALO's response. If AI Cloud falls back to an alternative processing pipeline, HALO context will not appear in the metadata.

Use Cases

Scenario

How to Implement

Personalize HALO responses based on CRM data

Set AI Cloud conversation variables with customer data; HALO automatically inherits matching context

Trigger UI actions from HALO

Set a context variable in HALO (e.g., show_feedback = true); consume it from AI Cloud's response metadata

Maintain user preferences across sessions

Set preferences in HALO context; AI Cloud persists them as conversation variables

Best Practices

  • Use consistent naming conventions — Context sync relies on matching variable names between systems. Establish a naming convention to avoid mismatches.

  • Combine with transient variables when needed — For one-time triggers (like UI events), use transient context variables in HALO to prevent unintended persistence.

  • Check the response source — If your integration depends on metadata, ensure you handle cases where AI Cloud falls back to non-HALO processing.