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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > Change Request Risk Agent

Change Request Risk Agent

Author: Venkata Sudhakar

Every deployment, configuration change, and infrastructure update at ShopMax India carries some risk. A database schema migration affecting the payment system is far riskier than a CSS update to the marketing website. An intelligent change risk agent can assess each change request and recommend the appropriate approval pathway - standard, expedited, or change advisory board.

This tutorial builds a Gemini ADK agent that evaluates a change request based on system criticality, change type, deployment window, and rollback feasibility, then assigns a risk rating and recommended approval process.

The below example shows a change request risk assessment agent for ShopMax India in a business context.

import asyncio
from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai import types

SYSTEM_CRITICALITY = {
    "Payment Gateway": 40,
    "Order Management": 35,
    "WMS": 30,
    "CRM": 20,
    "Marketing Website": 10,
    "Employee Portal": 10,
}

CHANGE_TYPE_RISK = {
    "database_migration": 30,
    "config_change": 15,
    "code_deployment": 20,
    "infrastructure_change": 25,
    "hotfix": 10,
    "ui_update": 5,
}

def assess_change_risk(system: str, change_type: str, rollback_possible: bool,
                       deployment_window: str, description: str) -> dict:
    sys_score = SYSTEM_CRITICALITY.get(system, 15)
    chg_score = CHANGE_TYPE_RISK.get(change_type, 15)
    rollback_score = 0 if rollback_possible else 20
    window_score = 0 if deployment_window == "maintenance" else 10
    total = sys_score + chg_score + rollback_score + window_score
    if total >= 70:
        risk = "HIGH"
        process = "Change Advisory Board (CAB) approval required. 48-hour notice."
    elif total >= 40:
        risk = "MEDIUM"
        process = "Manager + IT Lead approval. 24-hour notice. Deploy in maintenance window."
    else:
        risk = "LOW"
        process = "Standard approval. Team lead sign-off sufficient."
    return {
        "system": system,
        "change_type": change_type,
        "description": description,
        "rollback_possible": rollback_possible,
        "deployment_window": deployment_window,
        "risk_score": total,
        "risk_level": risk,
        "approval_process": process,
    }

change_tool = FunctionTool(func=assess_change_risk)

agent = LlmAgent(
    name="change_risk_agent",
    model="gemini-2.0-flash",
    description="Assesses IT change request risk for ShopMax India and recommends the approval pathway.",
    instruction=(
        "You are an IT change management agent for ShopMax India. "
        "When a change request is submitted, call assess_change_risk. "
        "State the risk level and required approval process clearly. "
        "If HIGH risk, list specific pre-change checks the team should complete."
    ),
    tools=[change_tool],
)

async def main():
    session_service = InMemorySessionService()
    runner = Runner(agent=agent, app_name="change_app", session_service=session_service)
    session = await session_service.create_session(app_name="change_app", user_id="it1")
    msg = types.Content(role="user", parts=[types.Part(
        text=(
            "Change request: System=Payment Gateway, Type=database_migration, "
            "Rollback=No, Window=business_hours, "
            "Description=Adding new columns to transactions table for UPI reference tracking"
        )
    )])
    async for event in runner.run_async(user_id="it1", session_id=session.id, new_message=msg):
        if event.is_final_response() and event.content:
            print(event.content.parts[0].text)

asyncio.run(main())

It gives the following output,

Change Risk Assessment - ShopMax India IT

System      : Payment Gateway
Change Type : Database Migration
Description : Adding new columns to transactions table for UPI reference tracking
Rollback    : NOT POSSIBLE
Window      : Business Hours

Risk Score  : 100 (System:40 + Change:30 + No-rollback:20 + Off-window:10)
Risk Level  : HIGH

Approval Required: Change Advisory Board (CAB) - 48-hour notice

Pre-change Checklist:
  1. Create full database backup before migration
  2. Test migration on staging environment first
  3. Schedule during 2-4 AM maintenance window (not business hours)
  4. Define rollback SQL script even if schema is additive
  5. Notify payment operations team before execution

ShopMax India can embed this agent into the Jira change request workflow so every ticket receives an automatic risk score before a human reviewer sees it. HIGH risk changes requiring CAB review are blocked from deployment until the approval chain is completed, reducing the risk of production incidents caused by insufficiently reviewed changes.

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