FAQ & Features
The IDXGenius Difference
Robust Workflows
IDXGenius supports any mortgage process or function that relies on heavy document processing.
Emerging Technologies
IDXGenius combines AI, machine learning, intelligent automation and business rules to accelerate accurate document-to-data extraction.
Eliminates Manual
Eliminates manual document interventions for cost reduction and productivity gains
Integrated with LOS
Offers fully automated indexing with an bi-directional LOS integration, eliminating the manual upload/download docs.
Accurate Results
Tested against thousands and thousands of files, IDXGenius drives even greater data accuracy, reducing risk at every step.
Time Savings
Expedites document-to-date processing while enabling your mortgage associates to focus on higher value tasks.
IDXGenius was initially built using traditional Optical Character Recognition (OCR) technology. Over time, it has evolved to incorporate advanced, industry-leading technology components such as AWS Textract and Google’s Tesseract. Additionally, IDXGenius leverages machine learning (ML) algorithms to learn and recognize variations in document types, continuously improving its accuracy and efficiency.
IDXGenius delivers an accuracy level of 95%+ for standard loan documents, ensuring reliable data extraction.
Yes, IDX, the tech component backing Indecomm’s Genius suite of technology and automation, was recognized as the “Best Innovation Using AI/ML” at the Amazon Web Services (AWS) AI Conclave in 2023.
Yes, IDXGenius has an ICE-certified integration with Encompass, facilitating a “Lights out” mode transfer of documents without requiring human intervention. Additionally, Indecomm’s SourceConnect (a part of Indecomm’s Genius tech stack) serves as a connectivity hub that can be configured to integrate with any other source system for data and document exchange. We assess the lender’s technology to provide tailored integration solutions. IDX is also a core component of Indecomm’s Genius tech stack which powers many of its own Genius technologies.
No, IDXGenius is a fully managed service integrated with the LOS (Encompass). Only the outputs (indexed documents) are returned to the eFolder. Indecomm manages any necessary updates or new document types, working directly with clients to ensure the service meets their needs.
IDXGenius supports a variety of loan products, including:
- Conventional
- FHA
- VA
- Non-QM (Non-Qualified Mortgage) products
However, non-QM products, which often contain a high number of non-standard documents (such as bank statements or DSCR), may result in higher manual exceptions. Indecomm will handle these exceptions and deliver the indexed files back into the LOS without requiring lender intervention.
IDXGenius is trained on a large dataset of mortgage documents, covering numerous variations. When new formats or variations arise, we update the model within a few days. Our team also tracks changes to standard forms, such as the 1003, 1008, LE, CDs, and tax returns, and integrates updates promptly. While we aim to handle accuracy monitoring entirely, we welcome client feedback to further enhance the system.
Over the past 8 years, we’ve continually refined our automation technology to enhance its effectiveness. After testing various solutions, we partnered with AWS, which significantly improved data extraction accuracy and allowed us to train models on a large dataset of mortgage-specific documents. Our dedicated automation team continuously works on optimizing components through ongoing improvement initiatives to ensure accuracy and reliability.
Over the past 8 years, we’ve continually refined our automation technology to enhance effectiveness. Partnering with AWS allowed us to achieve significant improvements in data extraction accuracy and train models on a large dataset of mortgage-specific documents. We also implement advanced techniques to boost accuracy and maintain a continuous improvement initiative, with a dedicated team optimizing our automation components regularly.
IDXGenius ensures accuracy using confidence scoring and multiple validation techniques:
- Auto-Correction: Algorithms analyze and adjust extracted data, especially for text fields.
- Formatting Validation: Errors in dates or numbers are auto-corrected using predefined data formats.
- Threshold Checks: Values like percentages are validated against limits (e.g., 0-100%).
- Generative AI: Large Language Models (LLMs) analyze documents, flagging fields below 90% confidence for review.
- Customizable Scores: Confidence levels can be adjusted to determine when human review is required.
When IDX detects low confidence or uncertainty, it flags the affected fields for review. Here’s how it works:
- Document Identification: Documents are first identified and categorized using machine learning.
- Data Extraction: IDX extracts data and applies various techniques like algorithms and Generative AI to validate the information.
- Data Certification: Fields with issues, such as low confidence scores or errors, are flagged for human review by our data quality team.
- Continuous Improvement: Our automation team continuously fine-tunes the models and updates the system to address future variations and improve extraction accuracy.
In this process, documents or fields with low confidence are handled by the data quality team, ensuring consistent refinement and accuracy.
