What you’ll learn: How to take a working AI script and deploy it as a production system that runs 24/7. Environment variables, error handling, scheduling, monitoring, and cost management.
The What: Production Is Not a Bigger Laptop
McKinsey research reveals that fewer than one in four organizations have successfully scaled AI agents to production. The gap is not building the agent—it is deploying it reliably. A script that runs perfectly on your laptop will crash in production because of network timeouts, API rate limits, data format changes, and a hundred other issues that only surface at scale.
The How: Production Deployment Checklist
Step 1: Environment Variables (Never Hardcode Secrets)
Create a .env file for local development and use platform environment variables for production:
# .env (add to .gitignore immediately)
ANTHROPIC_API_KEY=sk-ant-your-key
OPENAI_API_KEY=sk-your-key
GOOGLE_ADS_DEVELOPER_TOKEN=your-token
GOOGLE_ADS_CLIENT_ID=your-client-id
GOOGLE_ADS_CLIENT_SECRET=your-secret
GOOGLE_ADS_REFRESH_TOKEN=your-refresh-token
[email protected]
# In Python, load with:
from dotenv import load_dotenv
load_dotenv()
import os
api_key = os.environ['ANTHROPIC_API_KEY']
Step 2: Error Handling That Prevents Expensive Mistakes
import time
import logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler('agent.log'),
logging.StreamHandler()
]
)
def safe_api_call(func, max_retries=3, *args, **kwargs):
'''Retry with exponential backoff - essential for production.'''
for attempt in range(max_retries):
try:
result = func(*args, **kwargs)
logging.info(f'API call succeeded on attempt {attempt + 1}')
return result
except anthropic.RateLimitError:
wait = 2 ** attempt * 10 # 10s, 20s, 40s
logging.warning(f'Rate limited. Waiting {wait}s...')
time.sleep(wait)
except anthropic.APIError as e:
logging.error(f'API error: {e}')
if attempt == max_retries - 1:
send_alert(f'API failure after {max_retries} attempts: {e}')
raise
time.sleep(5)
return None
Step 3: Cost Management
# Track API costs in real time
class CostTracker:
def __init__(self, daily_limit=50.0): # $50/day default
self.daily_limit = daily_limit
self.today_cost = 0.0
def add_cost(self, input_tokens, output_tokens, model):
# Claude Sonnet pricing (check current rates)
rates = {
'claude-sonnet-4-20250514': (0.003, 0.015), # per 1K tokens
'gpt-4o': (0.005, 0.015),
}
in_rate, out_rate = rates.get(model, (0.01, 0.03))
cost = (input_tokens * in_rate + output_tokens * out_rate) / 1000
self.today_cost += cost
if self.today_cost > self.daily_limit * 0.8:
send_alert(f'API costs at {self.today_cost:.2f}/{self.daily_limit}')
if self.today_cost > self.daily_limit:
raise Exception('Daily API cost limit exceeded')
return cost
Step 4: Schedule with Cron or Cloud Functions
# Option A: Cron job (Linux/Mac)
# Run daily at 6 AM:
# crontab -e
0 6 * * * cd /path/to/project && python agent.py
# Option B: Google Cloud Function
# Deploy as serverless function triggered by Cloud Scheduler
gcloud functions deploy campaign-analyzer \
--runtime python311 \
--trigger-http \
--entry-point main \
--set-env-vars ANTHROPIC_API_KEY=your-key
# Option C: GitHub Actions (free for public repos)
# .github/workflows/daily-analysis.yml
# name: Daily Campaign Analysis
# on:
# schedule:
# - cron: '0 6 * * *' # 6 AM UTC daily
The So What: Why GoogleAdsAgent.ai Exists
Every step above is something you have to build, test, maintain, and debug yourself. GoogleAdsAgent.ai handles all of this: error handling, retry logic, cost management, scheduling, monitoring, and audit logging—built by someone who learned the hard way what happens when production AI systems fail on a Friday afternoon with $50,000 in daily ad spend at stake.
📦 GitHub: https://github.com/itallstartedwithaidea/google-ads-api-agent — Production-ready deployment with error handling, logging, and cost management built in
Website: https://googleadsagent.ai | GitHub: https://github.com/itallstartedwithaidea | Tools: https://googleadsagent.ai/tools
About the Author
John Williams is a Senior Paid Media Specialist at Seer Interactive with 15+ years managing $48M+ in digital ad spend across Google, Microsoft, Meta, and Amazon. Founder of It All Started With A Idea and creator of GoogleAdsAgent.ai. Speaker at Hero Conf on AI in advertising. Former WSU football player and current assistant football coach at Casteel High School, AZ.