Adobe ColdFusion includes a native, vendor‑neutral AI framework that lets you embed generative AI capabilities directly into CFML applications. The framework abstracts provider‑specific APIs behind a consistent set of services, so you can focus on building intelligent features instead of managing model integrations.

ColdFusion AI services are designed to scale with your application needs. You can start with simple prompt‑based interactions and progressively add memory, tools, orchestration, and document grounding as requirements evolve. Whether you are building lightweight AI helpers or enterprise‑grade AI agents, the same core architecture applies.

Use the sections to explore each part of the ColdFusion AI stack and understand when to use it.

Chat models provide stateless access to large language models (LLMs). Each request is handled independently, without retaining conversation history.

Use chat models for simple AI tasks such as text generation, summarization, translation, or one‑off question answering.

AI services build on chat models to support stateful, multi‑turn conversations. They add capabilities such as conversation memory, system messages, and consistent assistant behavior.

Use AI services when you are building conversational assistants, in‑application copilots, or AI agents that need to remember context and follow defined behavioral rules.

View Create a chat model for more information.

Model Context Protocol (MCP) enables AI services to integrate with external tools and systems using a standardized client‑server protocol. MCP tools can be hosted independently and shared across applications or teams.

Use MCP when you need distributed, reusable tools or deep integration with external enterprise systems.

View MCPs in ColdFusion for more information.

Vector stores enable semantic search by storing and querying embeddings generated from text or documents. Instead of relying on keyword matches, vector stores retrieve content based on meaning and similarity.

In ColdFusion, vector stores are used as the foundation for semantic search and retrieval‑augmented generation (RAG) workflows.

View Understand vector stores and embeddings in ColdFusion for more information.

Retrieval‑augmented generation (RAG) allows AI services to generate responses using your own documents and data rather than relying solely on model knowledge.

ColdFusion RAG combines embeddings, vector stores, and chat models into a configurable pipeline for building accurate, document‑grounded AI experiences such as knowledge assistants and help systems.

View Introduction to RAG in ColdFusion for more information.