We Pulled in the Sky’s Receipts

  • Storm data from NOAA and the National Weather Service isn’t sexy—but it’s accurate. We set up a feed to detect when hail, wind shear, or tornadic activity had crossed thresholds in high-roof-density zip codes.
  • Back in the day, this pipeline ran through Apache Airflow. These days, we’d use Mage—cleaner interface, faster deployments, less screaming. Either way, the job is the same: pull, parse, and prep the weather event data for execution.

We Automated the Physical World with Lob

  • Once the storm data hit, we matched the affected zones to residential address datasets. If your roof was in the blast radius, you got a postcard. Not a week later. Within 48 hours.
  • We used Lob—a postcard mailing API—to turn geospatial storm paths into targeted, printed mailers. Real ink, real paper, real fast. The kind of mail that lands before the tarps are even on.

Timing Isn’t a Metric—It’s Everything

  • Storm damage is a “first-in-wins” game. Homeowners aren’t loyal. They’re panicked. The contractor who gets in front of them first often gets the job, and that contractor? Buys shingles. So if you’re in the shingle business, speed is revenue.
  • This system turned a hurricane into a sales funnel. A windstorm into a logistics trigger. Weather was no longer random—it was monetized.

AWS Made It Scalable (and Sanely So)

  • Behind the scenes, we stitched together AWS services to make this not just fast, but repeatable:
  • AWS Lambda for event-driven processing (no more cron jobs, ever)
  • AWS S3 to store the raw NOAA data and cleaned up zip code maps
  • Amazon EventBridge to trigger workflows when thresholds were met
  • AWS Secrets Manager to keep API keys for Lob, Klaviyo, and HubSpot safe from human error

Customer Dashboards Made it Click

We plugged everything into dashboards using Klaviyo and HubSpot, so regional sales teams didn’t need to open a terminal to see what was happening.

They got:

  • Lead lists sorted by storm severity
  • Automated email sequences for contractors
  • Real-time views of postcard drops and homeowner responses

Now when a storm hits, the system kicks in—no Slack messages, no panic, just pure automation.

This is how you take a 100-year-old building product and give it 2025-level intelligence. Want to do the same for siding, windows, or solar? We’re already halfway there.