Overview
Comprehend is our advanced document classification and extraction engine. It can operate as a standalone application, integrate into existing systems, be deployed as SaaS, or function as a core component of our IDPS Automate platform.
Design Philosophy
Comprehend is built to be modular and deployment-agnostic, allowing seamless integration into your workflows without locking you into a specific implementation model. Deploy it on-premises for full control, run it in the cloud for scalability, or let us manage it entirely as a turnkey service.
We leverage state-of-the-art AI and ML techniques, combined with proven software engineering best practices, to ensure consistent, high-performance results at scale. Our approach is not tied to a single methodology—we blend multiple techniques to create tailored solutions that meet your exact requirements. This flexibility allows us to fine-tune performance for your unique document types, processing volumes, and operational needs.
How it Works
- Discovery Call – We review your environment, requirements, and document types.
- Solution Planning – We design a deployment strategy based on your goals and infrastructure.
- Model Training or Calibration – We either train models from scratch or fine-tune pre-trained models.
- Validation – Models are tested against sample production documents to ensure accuracy.
- Integration – The solution is embedded into your existing workflows (if applicable).
- Ongoing Optimization – We monitor and adjust as needed to address data drift, ensuring accuracy even as document formats change over time.
Accuracy & Reliability
Comprehend is a re-imagining of the proven technologies behind the processing of 10+ billion pages from over 2,000 document types in production—achieving 99%+ accuracy.
- If a document cannot be confidently classified, the system flags it instead of guessing.
- We track and analyze both unclassified and misclassified documents to improve accuracy quickly.
- Extraction follows the same intentional design, ensuring the most reliable field values possible.
Flexibility for Any Scale
We offer solutions for both small businesses (<1M pages/year) and large enterprises (>1B pages/year), with pricing aligned to actual needs—no paying for unnecessary features.
Core Options
- Classification Only – Identifies document type.
- Classification + Extraction – Identifies document type and extracts data fields.
Management Methods
- Managed – Continuous monitoring, error handling, and optimization. Ideal for large volumes, complex document types, or frequently changing formats.
- Unmanaged – Minimal oversight, suitable for low-volume, stable document types.
Deployment Formats
- .NET Library – Embed directly into your existing .NET application.
- Windows Application – Simplified interface for small businesses; best for ≤1M pages/year.
- Service (REST API) – Scalable API deployment with load-balancing support for high concurrency.
All deployment options share the same underlying engine, ensuring consistent accuracy and performance.
Pricing
Pricing is determined by:
- Number of document types
- Annual processing volume
- Deployment model and environment
We recommend a brief consultation to:
- Explore volume discounts for large-scale processing
- Identify the optimal configuration for your needs
- Provide a cost analysis comparing your current process to Comprehend