Google Agent

The Google Agent node is a specialized AI agent that uses Google's AI models (Gemini) with the unique ability to ground responses using Google Search. This enables the agent to access and incorporate current, real-time information from the web.

Overview

Use the Google Agent node when you need to:

Answer questions requiring current information
Research topics with up-to-date data
Generate content based on recent events
Verify or supplement information with web sources

Configuration

Field	Type	Required	Description
`model`	`string`	Yes	The Google AI model to use. Only Google/Gemini models are available.
`enableSearch`	`boolean`	No	When enabled, the model can search Google for current information. Default: `false`.
`agentPrompt`	`string`	Yes	The system prompt that defines the agent's role and instructions. Supports variable pills.
`inputString`	`string`	Yes	The user message or input for the agent to process. Supports variable pills.

Google Search Grounding

When Enable Google Search is turned on, the agent can:

Access real-time information from the web
Ground responses in current facts and data
Provide more accurate answers for time-sensitive queries
Cite or reference current sources

Note: Enabling Google Search may increase response time as the model performs web searches.

When to Enable Search

Use Case	Enable Search?
Current news or events	Yes
Real-time data (stock prices, weather)	Yes
Historical facts	Optional
Creative writing	No
Code generation	No
General knowledge	Optional

Loop Mode

The Google Agent node supports Loop Mode for batch processing:

Field	Type	Default	Description
`loopMode`	`boolean`	`false`	Enable to process each item in an array input separately.
`maxIterations`	`number`	`100`	Maximum number of iterations when loop mode is enabled.
`concurrency`	`number`	`1`	Number of parallel executions.
`onError`	`string`	`"stop"`	Error handling: `"stop"` or `"continue"`.

Inputs

The Google Agent node accepts inputs from connected upstream nodes. These inputs are available as variable pills that can be dragged into the Agent Description or Input fields.

Variable Pills

Variable pills allow you to reference outputs from other nodes:

Drag and drop pills from the "Available Inputs" section
Pills are automatically replaced with actual values at runtime
Use the {{variableName}} syntax for manual insertion

Output

Variable	Type	Description
`result`	`string`	The generated text response from the Google AI model.

When Loop Mode is enabled, the output is an array of strings, one for each processed input.

Example Use Cases

Current Events Research

Enable Search: Yes

Agent Description:

You are a research assistant. When asked about current events, provide accurate, up-to-date information with relevant context.

Input:

What are the latest developments in {{topic}}?

Company Information Lookup

Enable Search: Yes

Agent Description:

You are a business research assistant. Look up current information about companies and provide factual summaries.

Input:

Provide a summary of {{companyName}}, including recent news and current stock price if available.

Content Fact-Checking

Enable Search: Yes

Agent Description:

You are a fact-checker. Verify the claims in the provided content against current, reliable sources.

Input:

Fact-check the following article: {{articleContent}}

Knowledge Base Generation (No Search)

Enable Search: No

Agent Description:

You are a technical writer. Generate documentation based on the provided specifications.

Input:

Create user documentation for: {{featureSpec}}

Differences from Standard Agent

Feature	Agent	Google Agent
Model options	Multiple providers	Google models only
Google Search	Not available	Available (optional)
Real-time data	No	Yes (with search enabled)
Best for	General text generation	Tasks requiring current information

Best Practices

Enable search only when needed: Google Search adds latency. Only enable it for queries that benefit from real-time information.
Be specific about information needs: In your system prompt, clarify whether the agent should prioritize recency, accuracy, or comprehensiveness.
Consider search for verification: Even for topics you think are static, enabling search can help verify and update information.
Handle search limitations: The agent may not always find relevant search results. Include fallback instructions in your system prompt.
Monitor for time-sensitive content: When using search for time-sensitive queries, consider how quickly the information might become outdated.

Pricing Note

Google Search grounding may affect credit usage. Check the Credits and Costs page for current pricing information.

Google Agent

On this page