	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > ADK with Cloud Scheduler

ADK with Cloud Scheduler

Author: Venkata Sudhakar

Not every agent workflow needs a human trigger. Daily sales summaries, morning inventory alerts, weekly performance reports, and end-of-month reconciliation tasks all run on fixed schedules. Google Cloud Scheduler is a fully managed cron service that sends HTTP requests to a Cloud Run endpoint at a configured time, making it straightforward to run ADK agents on autopilot.

ShopMax India uses Cloud Scheduler to run three recurring agent tasks: a 7am daily stock alert that flags items below reorder level across all warehouses, a 6pm daily sales summary sent to the management team, and a Monday morning weekly performance report covering all cities. All three are ADK agents triggered by scheduled HTTP calls to Cloud Run endpoints.

The below example shows a Cloud Run service that hosts a scheduled ADK agent to generate a ShopMax India daily stock alert report.

from flask import Flask, request, jsonify
from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool
from google.adk.runners import InMemoryRunner
from google.adk.sessions import InMemorySessionService
from datetime import datetime

app = Flask(__name__)

def get_low_stock_items() -> list:
    """Fetch items below reorder level from ShopMax India inventory."""
    # In production, query Databricks or BigQuery
    return [
        {"sku": "SM-PHONE-OP12", "city": "Delhi", "stock": 8, "reorder_level": 25},
        {"sku": "SM-PHONE-S24", "city": "Pune", "stock": 3, "reorder_level": 20},
        {"sku": "SM-AUDIO-BOAT", "city": "Chennai", "stock": 12, "reorder_level": 30}
    ]

def send_alert_email(report: str, recipients: list) -> dict:
    """Send the stock alert report by email."""
    print(f"Email sent to {recipients}:\n{report}")
    return {"status": "sent", "recipients": len(recipients)}

stock_alert_agent = LlmAgent(
    name="stock_alert_agent",
    model="gemini-2.0-flash",
    instruction="""You are the ShopMax India automated stock monitoring agent.
When triggered, call get_low_stock_items to retrieve current stock data.
Produce a clear alert report with a summary table and recommended actions.
Then call send_alert_email with the report and recipients list.
Recipients: ['operations@shopmax.in', 'procurement@shopmax.in']""",
    tools=[FunctionTool(func=get_low_stock_items), FunctionTool(func=send_alert_email)]
)

session_service = InMemorySessionService()
runner = InMemoryRunner(agent=stock_alert_agent, session_service=session_service)

@app.route("/scheduled/stock-alert", methods=["POST", "GET"])
def run_stock_alert():
    run_time = datetime.utcnow().isoformat()
    print(f"[Scheduler] Stock alert triggered at {run_time}")
    session = session_service.create_session(
        app_name="shopmax_scheduler", user_id=f"scheduler_{run_time}"
    )
    result = runner.run(
        user_id=f"scheduler_{run_time}",
        session_id=session.id,
        new_message="Run the daily stock alert check for ShopMax India."
    )
    output = ""
    for event in result:
        if hasattr(event, "content") and event.content:
            for part in event.content.parts:
                if hasattr(part, "text") and part.text:
                    output += part.text
    return jsonify({"status": "completed", "run_time": run_time, "output_length": len(output)}), 200

if __name__ == "__main__":
    app.run(host="0.0.0.0", port=8080)

It gives the following output,

[Scheduler] Stock alert triggered at 2026-04-06T01:30:00 (07:00 IST)

SHOPMAX INDIA - Daily Stock Alert Report
Generated: 06 April 2026, 07:00 IST

ITEMS BELOW REORDER LEVEL:
  SKU            | City    | Stock | Reorder | Shortfall
  SM-PHONE-OP12  | Delhi   |   8   |   25    |   -17
  SM-PHONE-S24   | Pune    |   3   |   20    |   -17
  SM-AUDIO-BOAT  | Chennai |  12   |   30    |   -18

RECOMMENDED ACTIONS:
  1. Raise emergency PO for SM-PHONE-S24 (Pune) - critically low at 3 units
  2. Transfer 10 SM-PHONE-OP12 units from Mumbai to Delhi warehouse
  3. Place standard reorder for SM-AUDIO-BOAT (Chennai)

Email sent to ['[email protected]', '[email protected]']

The below example shows the gcloud CLI commands to deploy the Cloud Run service and create the Cloud Scheduler job that triggers it daily at 7am IST.

# Deploy the scheduled agent as a Cloud Run service
gcloud run deploy shopmax-scheduled-agents \
  --image gcr.io/shopmax-india/scheduled-agents:latest \
  --platform managed \
  --region asia-south1 \
  --no-allow-unauthenticated

# Create Cloud Scheduler job: daily stock alert at 7am IST (01:30 UTC)
gcloud scheduler jobs create http shopmax-daily-stock-alert \
  --location=asia-south1 \
  --schedule="30 1 * * *" \
  --uri="https://shopmax-scheduled-agents-xyz.run.app/scheduled/stock-alert" \
  --http-method=POST \
  --oidc-service-account-email=scheduler-sa@shopmax-india.iam.gserviceaccount.com \
  --time-zone="Asia/Kolkata"

# Create weekly Monday 8am IST performance report job
gcloud scheduler jobs create http shopmax-weekly-report \
  --location=asia-south1 \
  --schedule="30 2 * * 1" \
  --uri="https://shopmax-scheduled-agents-xyz.run.app/scheduled/weekly-report" \
  --http-method=POST \
  --oidc-service-account-email=scheduler-sa@shopmax-india.iam.gserviceaccount.com \
  --time-zone="Asia/Kolkata"

# List active jobs
gcloud scheduler jobs list --location=asia-south1

It gives the following output,

Deploying Cloud Run service... done
Service URL: https://shopmax-scheduled-agents-xyz.run.app

Creating scheduler job: shopmax-daily-stock-alert... done
Schedule: 30 1 * * * (daily 07:00 IST)
Target: POST /scheduled/stock-alert

Creating scheduler job: shopmax-weekly-report... done
Schedule: 30 2 * * 1 (Monday 08:00 IST)
Target: POST /scheduled/weekly-report

ID                         LOCATION      SCHEDULE       STATUS
shopmax-daily-stock-alert  asia-south1   30 1 * * *     ENABLED
shopmax-weekly-report      asia-south1   30 2 * * 1     ENABLED

Cloud Scheduler with ADK agents replaces brittle cron scripts and manual reporting workflows at ShopMax India. Each scheduled job runs a fully capable Gemini-powered agent that can query live data, reason about results, generate narrative summaries, and dispatch notifications - all automatically, on time, every day, without any human in the loop.

Send your comments, suggestions or queries regarding this site to [email protected].