Real T Solutions

Client

Real T Solutions

Location

Las Vegas, USA

Engagement

AI Automation and Orchestration for Property Intelligence

Technologies & Tools Used

Django
Django
Playwright
Playwright
Python
Python
PostgreSQL
PostgreSQL
Selenium
Selenium
AWS
AWS
Airflow
Airflow

From manual county research to intelligent automation at county scale.

Real T Solutions works in title research and property verification, running large-scale research across assessor, recorder, tax, and court records published by public authorities.

These records are spread across thousands of county websites, each with its own layout, navigation flow, and data format.

Real T Solutions transforms this fragmented public-record data into verified property intelligence at county-level scale.

Building a reliable research platform across 1,000+ jurisdictions.

Manual navigation across county websites was slow, error-prone, and difficult to maintain at scale.

LLM-based automation could infer website structures, but without backend orchestration, scraping runs lasting one to two hours or more could time out or fail silently.

A single county failure often forced the entire order to be processed again, increasing turnaround time and infrastructure cost.

Spark Eighteen helped Real T Solutions replace brittle, manually maintained county scrapers with an LLM-driven, schema-based research platform built for reliability across more than 1,000 jurisdictions.

The Challenge

Real T Solutions needed to automate research across fragmented county websites while avoiding full pipeline reruns and reducing failures caused by frequent website changes.

1,000+ Fragmented Counties

Every county website had unique structures, layouts, and navigation flows to scrape.

Full Re-Runs on Failure

One failed data source forced the entire pipeline to restart, even when other steps succeeded.

Brittle Scrapers

Frequent UI and DOM changes broke existing scraping logic without warning.

The Solution

Spark Eighteen used LLMs to read county websites and infer DOM structures and extraction rules.

The platform generates a unified JSON schema that drives Playwright execution, eliminating the need for county-specific scraper code.

A Django orchestration layer and long-running Airflow DAGs provide task-level state, granular retries, and selective re-scraping that preserves completed work.

The Results

Real T Solutions now runs title research across hundreds of counties with minimal human intervention.

Failures recover without requiring completed work to be processed again.

100+

Counties researched end-to-end with minimal human intervention

Business Outcomes

Manual effort cut - Large-scale title research workflows automated with minimal human intervention.

Accuracy up - Standardized outputs increased data reliability across jurisdictions.

Faster county onboarding - Reusable pipeline patterns cut engineering effort per new county.

Safe partial retries - Failures reprocess selectively without duplicating completed work.

Observability built in - Task-level visibility sped up issue diagnosis and resolution.

Scalable automation - Fragmented public record workflows became a repeatable, high-accuracy pipeline.

Spark Eighteen helped us move from fragile, scraper-heavy automation to a reliable and scalable title research platform. Their LLM-driven approach combined with strong backend orchestration gave us the observability, control, and resilience we were missing. We can now operate across hundreds of counties with far less manual effort and significantly improved turnaround times.

—Team Real T Solutions

View Related Projects
connect-hero-thumbnail

ConnectHEOR

Dayatani

Dayatani

Fitbudd

FitBudd

© 2026 All rights reserved •

Spark Eighteen Lifestyle Pvt. Ltd.