Academy
⌘K
Getting Started
  • Platform Overview
  • AWS Marketplace Setup
  • Quickstart Guide
Connections
  • Overview
Workflow Builder
  • Overview
  • Nodes Catalog
Helix AI Assistant
  • Overview
Mapping Engine
  • Overview
Workflow Library
  • Overview
  • HL7
  • FHIR
  • Insurance
  • Agentic
  • Mapping
Security & Infrastructure
  • Architecture
  • HIPAA Compliance
  • Performance & Scalability
← Back to ConnectHealth
Get Started
Back to Nodes
AI Agent1 input → 1 output

AI Model

payload (any)aiagent (object)AI AgentAI Modelpayload (any)aiagent.provider (string)aiagent.model (string)Data flowProperty

Configuration node that defines AI model settings and credentials for use with AI Agent nodes. Supports multiple AI providers.

Configuration

This is a configuration node—it doesn't process messages directly. Configure it once and reference from AI Agent nodes.

Name string
Descriptive name for this configuration (e.g., "Production Claude", "Dev Nova Lite")
Provider select
AI provider to use:
  • Vertex AI: Google's AI platform (Gemini models)
  • AWS Bedrock: Amazon's managed AI service (Claude, Titan, Llama)
Temperature number (0-2)
Controls randomness. 0 = deterministic, 1 = balanced, 2 = very creative. Default: 0.7
Max Tokens number
Maximum tokens in AI response. Higher = longer responses but more cost. Default: 1000

Provider: Vertex AI

Google Cloud's managed AI platform featuring Gemini models.

Project ID string
Your Google Cloud project ID
Location select
GCP region (e.g., us-central1, europe-west1)
Service Account JSON credentials
Paste your service account JSON key. Requires Vertex AI API enabled.
Model select
Gemini models: 2.0 Flash, 1.5 Pro, 1.5 Flash

Provider: AWS Bedrock

Amazon's managed service with models from Anthropic, Amazon, Meta.

AWS Region select
Region where Bedrock is enabled (e.g., us-east-1, us-west-2)
Authentication select
Choose method:
  • Environment Variables: Uses AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY (recommended)
  • Access Key + Secret: Enter credentials in node config
  • API Key: For custom implementations
AWS Access Key ID credentials
Required when using "Access Key + Secret" authentication
AWS Secret Access Key credentials
Required when using "Access Key + Secret" authentication
Model text with autocomplete
Models: Anthropic Claude, Amazon Titan, Meta Llama. Requires inference profile prefix (e.g., us.)

Outputs

aiagent object
The AI settings for the AI Agent and Guardrails nodes: provider, model, temperature, maxTokens, and the region or project settings for the provider.
aiagent.credentialRef string
The id of this node. The AI Agent and Guardrails nodes use it to look up this node's credentials when they call the provider. No key, access key or service-account JSON is ever put on the message, so none can show up in the Debug panel, a Catch node or an HTTP response.

Details

This configuration node stores shared AI settings used across AI Agent nodes in your flows.

Deprecated: reading secrets from the message

Reading msg.aiagent.apiKey (or accessKeyId, secretAccessKey, credentials) in a downstream node still returns this node's credential, but it is deprecated and logs a warning once per deploy. The value is hidden from the Debug panel. Move any flow that reads it to the AI Agent node, which uses credentialRef instead.

Cost Optimization

  • Use cheaper models (Amazon Nova Lite, Gemini Flash) for simple tasks
  • Reserve expensive models for complex reasoning
  • Lower max_tokens for shorter responses

Security Best Practices

  • Never commit API keys to version control
  • Use environment variables for production
  • Rotate keys regularly

Examples

Multiple Configurations: Create separate configs for different use cases:

  • "Production Bot" - Stable model, lower temp (0.5)
  • "Creative Assistant" - Advanced model, higher temp (0.9)
  • "Cost-Effective" - Cheap model for high-volume tasks