OneDigital Transforms Quoting Intake with SortSpoke
By SortSpoke
“What impressed us most about SortSpoke was its ability to start delivering value immediately without extensive setup or template creation”
Kan Zhu, Director of Data & Analytics, Teranet
TL:DR
250,000 Documents Processed at 97% Time Savings with Zero IT Involvement
Manual document processing has always been a major headache for data teams that need to unlock valuable data from historical records, until now.
- Teranet, an international leader in land registration owned by OMERS, needed to extract valuable data from 250,000 historical Ontario condo declarations and Quebec land transfer documents
- Traditional data extraction methods and template-based solutions weren't economically viable due to document variety, multiple formats, and bilingual content
- SortSpoke began learning after just 15 documents on the first day with no templates required
- The solution delivered production-ready, high-quality data from the very first document
- Processing time was consistently under 20 seconds per document, representing a 97% time reduction compared to manual extraction
- Zero maintenance costs and no IT consultants needed as SMEs could manage the solution themselves, resulting in an estimated 85% cost savings over traditional template-based solutions
Teranet's experience demonstrates how SortSpoke's human-in-the-loop AI approach can make previously inaccessible document data economically viable to extract and utilize.
By combining flexible AI extraction that works without templates and human verification that ensures quality, SortSpoke enabled Teranet to unlock the value of 250,000 historical land registry documents quickly and cost-effectively.
"This project would have been prohibitively expensive using traditional approaches," concludes a Teranet executive. "SortSpoke made it possible to extract this valuable data efficiently while maintaining the high quality standards essential for our analytical and modeling purposes. The speed of implementation and lack of ongoing maintenance costs have made this a particularly successful technology investment."
Kan Zhu, Director, Data & Analytics, Teranet
Teranet's Challenges
Valuable Data Trapped in Diverse Historical Documents
Teranet, an international leader in land registration services owned by OMERS, possessed an extensive library of approximately 250,000 historical Ontario condo declarations and Quebec land transfer documents.
The data contained within these documents—including sale prices, property descriptions, and transaction dates—was extremely valuable for Canadian mortgage lenders for modeling, forecasting, and analysis.
However, extracting this data presented significant challenges:
- The documents originated from hundreds of different land registry offices across two provinces
- Formats and layouts varied extensively with no standardization
- Documents were in both English and French
- Data points appeared in various formats—tables, paragraphs, or single values scattered throughout the documents
- Some data spanned across multiple pages
"The value of the data was compelling, but both manual extraction and existing template-based technologies were economically unfeasible given the volume and variety of documents," explains a senior executive at Teranet.
"We needed a solution that could handle this diversity while delivering high-quality data for our database, analytics, and predictive modeling."We couldn’t get the documents summarized fast enough to get quotes delivered."
— Kan Zhu, Director, Data & Analytics Teranet
The Solution
SortSpoke's Flexible AI Extraction Without Templates
After evaluating several options, Teranet selected SortSpoke's AI-powered document processing platform to extract data from their vast document library.
SortSpoke's approach provided several key advantages that addressed Teranet's unique challenges:
✅ Rapid learning without templates: SortSpoke began learning after processing just 15 documents on the first day of implementation, eliminating the need for rigid templates that wouldn't work across varied document formats.
✅ Production-ready from day one: High-quality data extraction was achieved from the very first document, allowing Teranet to start building their database immediately.
✅ Adaptability to document variation: The platform easily handled tables, single values, sentences, and paragraphs that bridged pages, adapting on the fly to new land registry document formats.
✅ Bilingual capability: SortSpoke processed both English and French documents with equal efficiency.
✅ Human-in-the-loop validation: Every field was verified by a subject matter expert, ensuring data accuracy while maintaining the efficiency of AI-powered extraction.
"What impressed us most about SortSpoke was its ability to start delivering value immediately without extensive setup or template creation," notes a Teranet project manager. "The combination of AI extraction with human verification gave us confidence in the data quality while dramatically accelerating the process."
— Kan Zhu, Director, Data & Analytics, Teranet
.
implementation
Minimal Setup, Maximum Control
The implementation process was remarkably straightforward, especially given the complexity of the document set:
📉 Initial training: SortSpoke's team worked with Teranet to process an initial set of 15 documents, allowing the system to begin learning.
📈 Same-day results: Within hours of the initial training, the system was already extracting data from production documents
⏳User-controlled refinement: As Teranet's subject matter experts verified data, the system continuously improved its extraction accuracy.
🔮 Ongoing adaptability: When new document formats were encountered, SortSpoke adapted without requiring IT intervention or consultant assistance.
"The implementation was unlike anything we'd experienced with previous technology solutions," recalls a Teranet team member. "There was no lengthy setup period or extensive IT involvement. Our own subject matter experts were able to manage the entire process."
— Data Team Manager, Teranet
.
Results
Consistent Speed, Quality, and Cost-Effectiveness
Teranet quickly realized significant benefits from implementing SortSpoke:
Consistent processing speed: Documents were processed in under 20 seconds each, enabling rapid data extraction from the entire 250,000 document library
High-quality data: Human verification of every field ensured the accuracy needed for database building and analytics
Zero maintenance costs: No IT consultants were required as the system was fully managed by Teranet's own subject matter experts
Adaptability to new documents: As new land registry documents arrived, SortSpoke handled them without requiring reconfiguration
Bilingual processing: Both English and French documents were processed with equal efficiency
The solution also provided the flexibility to extract various data formats—whether in tables, paragraphs, or as single values—giving Teranet complete access to all the valuable information contained in their document library.
“Looking ahead, our goal for SortSpoke is to make it a building block for total process automation.”
— Beata Rogala, COO, Centro Benefits Research
.
