Conversational AI Cloud

Conversational AI Cloud (CAIC) is CM.com's platform for building and managing AI-powered customer conversations. It sits between your customers and your knowledge and logic — interpreting what users say, selecting the right response, and delivering it across channels.

CAIC is designed for teams that need control over how their bot behaves: what it recognises, what it answers, how it guides users through complex flows, and when it hands off to a human or an AI agent. It supports both rule-based and AI-driven approaches, and integrates directly with HALO for agentic AI capabilities.

What CAIC does

When a customer sends a message, CAIC:

1. Passes the message through its NLU pipeline to determine intent and find the best match

2. Returns the matched content — an article answer, an event output, or a dialog flow

3. Syncs conversation state across turns so context carries forward

4. Logs the interaction for analytics and continuous improvement

Key concepts

Projects

A project is the container for everything in CAIC — content, configuration, analytics, and publication settings. Each project typically represents a single bot or customer experience. Projects have a Staging environment for testing and a Production environment for live, customer-facing traffic.

Content types

CAIC has three content types that define what gets returned to users:

NLU pipeline

The NLU (Natural Language Understanding) pipeline determines how CAIC interprets user messages. It runs a sequence of configurable recognition engines — AI Cloud's built-in intent and rule-based recognition, and optionally HALO's agentic AI — and returns the first match. See Hybrid NLU for configuration details.

Configuration

Project configuration covers:

• Context variables — static values supplied by the client at the start of a conversation (e.g. customer tier, device type)

• Conversation variables — dynamic values that are set and updated during a conversation and persist across turns

• Metadata — structured data returned to the client in the API response for triggering UI actions or external logic

• Hybrid NLU — the NLU pipeline configuration

HALO integration

CAIC integrates with HALO, CM.com's agentic AI platform. HALO can be added as a processing stage in the NLU pipeline, called as an agent or tool output from any content item, and accessed for analytics. See HALO Integration for the full overview.

Analytics

CAIC's analytics dashboards provide insight into conversation volumes, recognition performance, dialog flow completion, customer feedback, and more. See CAIC Analytics.

Publication

Changes made in CAIC are saved as drafts and must be published to go live. CAIC supports separate Staging and Production environments. See CAIC Publication.

Where to start