Source
yaml
id: ai-agent-calling-flows
namespace: tutorial
description: AI Agent
inputs:
- id: use_case
type: SELECT
displayName: Select Your Orchestration Use Case
defaults: Just Exploring
values:
- Business Automation
- Business Processes
- Data Engineering Pipeline
- Data Warehouse and Analytics
- Infrastructure Automation
- Microservices and APIs
- Just Exploring
tasks:
- id: agent
type: io.kestra.plugin.ai.agent.AIAgent
prompt: |
Execute a flow that best matches the {{inputs.use_case}} use case selected by the user. Use the following mapping of use cases to flow ids:
- Business Automation: business-automation
- Business Processes: business-processes
- Data Engineering Pipeline: data-engineering-pipeline
- Data Warehouse and Analytics: dwh-and-analytics
- Infrastructure Automation: infrastructure-automation
- Microservices and APIs: microservices-and-apis
- Just Exploring: hello-world
Remember that all those flows are in the tutorial namespace.
Return only the Execution URI with no extra characters - the structure of URL is {{ kestra.url ?? 'http://localhost:8080/'}}ui/<tenantId>/executions/<namespace>/<flowId>/<id>
provider:
type: io.kestra.plugin.ai.provider.GoogleGemini
modelName: gemini-2.5-flash
apiKey: "{{ secret('GEMINI_API_KEY') }}"
tools:
- type: io.kestra.plugin.ai.tool.KestraFlow
- id: uri
type: io.kestra.plugin.core.debug.Return
format: "{{ (kestra.url ?? 'http://localhost:8080/') ~ 'ui/' ~
fromJson((outputs.agent.toolExecutions | first).result).tenantId ~
'/executions/tutorial/' ~ fromJson((outputs.agent.toolExecutions |
first).result).flowId ~ '/' ~ fromJson((outputs.agent.toolExecutions |
first).result).id }}"
About this blueprint
AI
This blueprint demonstrates how to use an AI agent to select and execute the most relevant automation based on a user’s stated intent or use case.
It shows how to:
- Capture high-level user intent such as business automation, data engineering, analytics, or infrastructure operations.
- Delegate decision-making to an AI agent that maps intent to the most appropriate predefined workflow.
- Automatically trigger the selected automation without hardcoded conditional logic.
- Use AI tool-calling to invoke orchestration actions programmatically.
- Return a direct execution URL so users can immediately inspect the result of the automated decision.
This pattern is commonly used in AI-driven orchestration, self-service automation platforms, internal developer portals, and agent-based systems where users should not need to know which workflow to run.
By allowing an AI agent to route requests to the correct automation, this approach reduces cognitive load, simplifies onboarding, and enables intent-based execution across complex automation catalogs.
More Related Blueprints