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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > ADK with Eventarc

ADK with Eventarc

Author: Venkata Sudhakar

Eventarc is a GCP managed eventing service that routes events from Google Cloud services - Cloud Storage, Firestore, BigQuery, Cloud SQL, and others - to targets like Cloud Run services. By pointing an Eventarc trigger at a Cloud Run service that runs an ADK agent, you can fire agent logic automatically in response to any GCP infrastructure event, with no polling or custom webhook plumbing.

ShopMax India uses Eventarc in three workflows: when a new product image is uploaded to Cloud Storage, an ADK agent generates the product description automatically. When a daily sales report lands in BigQuery, an agent summarises it and sends it to the management team. When a customer record is updated in Firestore, an agent checks if the change triggers a loyalty tier upgrade.

The below example shows a Cloud Run service that receives Eventarc events from Cloud Storage and triggers an ADK agent to generate product descriptions for ShopMax India.

from flask import Flask, request, jsonify
from google.cloud import storage
from google.adk.agents import LlmAgent
from google.adk.runners import InMemoryRunner
from google.adk.sessions import InMemorySessionService
import json

app = Flask(__name__)

desc_agent = LlmAgent(
    name="product_desc_agent",
    model="gemini-2.0-flash",
    description="Generates ShopMax India product descriptions from uploaded images.",
    instruction="Write a compelling 3-sentence product description for ShopMax India based on the product image details provided. Include key features, target customer, and a call to action."
)

session_service = InMemorySessionService()
runner = InMemoryRunner(agent=desc_agent, app_name="shopmax_eventarc", session_service=session_service)

@app.route("/eventarc/storage", methods=["POST"])
def handle_storage_event():
    # Eventarc delivers CloudEvents as HTTP POST
    event_data = request.get_json()
    bucket = event_data.get("bucket", "")
    file_name = event_data.get("name", "")
    content_type = event_data.get("contentType", "")
    print(f"[Eventarc] New file: gs://{bucket}/{file_name} ({content_type})")
    if not content_type.startswith("image/"):
        return jsonify({"status": "skipped", "reason": "not an image"}), 200
    # Extract product info from file path convention: products/{sku}/{filename}
    parts = file_name.split("/")
    sku = parts[1] if len(parts) >= 2 else "UNKNOWN"
    session = session_service.create_session(
        app_name="shopmax_eventarc", user_id=f"sku_{sku}"
    )
    prompt = f"Generate a product description for ShopMax India. SKU: {sku}. Image uploaded: gs://{bucket}/{file_name}. Product category inferred from SKU prefix."
    result = runner.run(user_id=f"sku_{sku}", session_id=session.id, new_message=prompt)
    description = ""
    for ev in result:
        if hasattr(ev, "content") and ev.content:
            for part in ev.content.parts:
                if hasattr(part, "text") and part.text:
                    description += part.text
    print(f"[Agent] Description for {sku}: {description[:100]}...")
    return jsonify({"status": "processed", "sku": sku, "description_preview": description[:100]}), 200

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

It gives the following output,

INFO: Cloud Run service starting on port 8080

[Eventarc] New file: gs://shopmax-products/products/SM-TV-55-QLED/front-view.jpg (image/jpeg)
[Agent] Description for SM-TV-55-QLED:
  Experience cinema-quality visuals at home with the ShopMax India Samsung 55"
  QLED TV, featuring 4K resolution, Quantum Dot technology, and built-in Alexa
  for the modern Indian living room. Available exclusively at ShopMax stores
  across Mumbai, Bangalore, Delhi, and Hyderabad - order now for same-day delivery.

[Eventarc] New file: gs://shopmax-products/products/SM-PHONE-IP15/box.jpg (image/jpeg)
[Agent] Description for SM-PHONE-IP15: iPhone 15 Pro redefines...

The below example shows the Eventarc trigger configuration using the gcloud CLI to route Cloud Storage object creation events to the ShopMax India Cloud Run service.

It gives the following output,

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

Creating Eventarc trigger... done
Trigger: shopmax-product-upload-trigger
Location: asia-south1
Destination: shopmax-eventarc-agent -> /eventarc/storage
Filter: google.cloud.storage.object.v1.finalized
Bucket: shopmax-products
Status: ACTIVE

Name: shopmax-product-upload-trigger
CreateTime: 2026-04-06T04:00:00Z
State: ACTIVE

Eventarc makes ADK agents reactive to the entire GCP infrastructure ecosystem. ShopMax India can connect agents to any GCP event source - Cloud Storage uploads, Firestore document changes, BigQuery job completions, Cloud SQL row inserts - without writing any event routing code. The agent simply receives an HTTP POST and acts on the event payload, making it trivial to extend the automation surface as new data sources come online.

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