Coforge implemented an AI-based solution to extract mortgage title data from unstructured documents and ingest it into our SmartProp agentic workflow platform to reduce processing time, improve accuracy and reduce cost.
The Challenge
The client was creating title reports manually, which required as much as 25 minutes of processing for each document included in the report. Establishing a clear chain of title can require analyzing dozens of public documents for each loan, making the process slow and cumbersome.
Manually extracting data and typing it into the report introduced errors and re-work, and made it difficult to scale the process during spikes in volume.
The client needed a way to increase speed and reduce errors in the process of generating title reports.
Our Approach
Coforge provided a comprehensive solution built on the SmartProp workflow platform, including an agentic AI-based data extraction and ingestion solution with intelligent reviews and human oversight.
Agentic Data Extraction
Specialized AI agents process public records, identify relevant data from unstructured documents, and ingest it into the SmartProp workflow platform.
Report Creation
Title reports were built using the LangGraph framework along with Microsoft OpenAI and MS Document Intelligence to interpret the data and autonomously populate it into the reports.
Report Review
The reports were validated through human in‑the‑loop (HITL) review, highlighting key fields and accelerating validation through assisted copy‑paste into the SmartProp platform.

Impact to Date
Up to 40%
Up to 87%
