Run dotAI on the AI provider you choose

Description

dotAI Multi-Provider Support lets developers and platform teams run the AI built into dotCMS on the provider they choose, so AI can be adopted inside the boundaries an organization already approves.

You configure the provider once in the dotAI app, in a single Provider Config field with three sections - chat, embeddings, and image - each naming its own provider and model. Every dotAI capability then runs on that provider: writing assistance in the block editor, AI workflow actions, automatic tagging, semantic search, and image generation. Seven providers are supported, and existing OpenAI setups keep working unchanged.

Supported providers: Azure OpenAI, AWS Bedrock, and Google Vertex AI for running inside your own cloud boundary; Anthropic and Google AI Gemini for direct API-key access with no cloud project; OpenAI; and OpenRouter, an aggregator that reaches hundreds of models across families through one API key.

Benefits

For Developers & Platform Teams

• Select the AI provider in configuration - no code changes - and swap providers or models by editing the dotAI app.

• Run AI inside the cloud you already use: Azure, AWS, or Google Cloud, including AWS auth through the default credential chain (environment, profile, or IRSA).

• Route AI traffic through an internal gateway with an endpoint override; provider credentials are redacted in logs.

For Content & Marketing Teams

• Keep using the same AI writing, tagging, and semantic search in the block editor - now running on an approved provider.

• Turn on AI features that were previously off-limits because of provider constraints.

• Use the best model for each type of job, search, images, content, and control costs by using lower cost models for less complex jobs.

Use Cases

Run AI inside an approved cloud
A financial-services organization that can only use AI hosted in its contracted cloud points dotAI at Azure OpenAI. Content teams keep using AI writing, tagging, and search - now inside the boundary the organization already approved.

Turn on AI without a cloud contract
A team with no cloud-AI agreement pastes a single Anthropic or Google AI Studio key into the chat section, and dotAI is live on Claude or Gemini in minutes - no GCP project, no AWS roles.

Keep model options open
A platform team configures OpenRouter once and can point chat at any of hundreds of models, changing models by editing a config value, all under one bill.

Mix providers by capability
Run chat generation on one provider and semantic-search embeddings on another by setting each section independently.

Resources

• Feature Blog Post

• Documentation