Hyperscience R36 Release
Automate More Use Cases
Classify text, emails, & unstructured documents by user intent, sentiment, or topic—letting ML technology find and extract the relevant data.
Scale The Use Of AI
Train better ML models faster with an ML-guided user experience that flags potential errors and anomalies during the training data labeling process.
Achieve End-To-End Processing
Reduce average handling time with an improved workflow engine that supports higher-quality decisions with required and dependent fields.
Text Classification
Reduce costs and accelerate operations by automating the classification and routing of unstructured documents—without sacrificing accuracy.
Classify any text input, including documents, emails, or raw text.
Train classification models based on intent, sentiment or topic without involving technical resources.
Learn more
Text Classification
Reduce costs and accelerate operations by automating the classification and routing of unstructured documents—without sacrificing accuracy.
Classify any text input, including documents, emails, or raw text.
Train classification models based on intent, sentiment or topic without involving technical resources.
Learn moreUnstructured Extraction
Improve the efficiency of complex processes that rely on unstructured data. Extract valuable insights—such as co-signers from deeds, or names from divorce decrees—to make better decisions in less time.
Unlock valuable insights trapped in unstructured documents.
Easily train models with your own data to extract points of interest from a document or use one of our out-of-the-box models to extract entities like name, address, or organization.
Learn More
Unstructured Extraction
Improve the efficiency of complex processes that rely on unstructured data. Extract valuable insights—such as co-signers from deeds, or names from divorce decrees—to make better decisions in less time.
Unlock valuable insights trapped in unstructured documents.
Easily train models with your own data to extract points of interest from a document or use one of our out-of-the-box models to extract entities like name, address, or organization.
Learn MoreLabeling Anomaly Detection
Get more from your intelligent automation investment. Scale the use of machine learning across your organization by building more robust ML models faster—without technical resources.
Teach AI to find the “needle in a haystack” during model training, where it highlights fields that aren’t consistent with similar documents.
Use an ML-guided interface to train higher-quality ML models in less time and with fewer samples.
Learn More
Labeling Anomaly Detection
Get more from your intelligent automation investment. Scale the use of machine learning across your organization by building more robust ML models faster—without technical resources.
Teach AI to find the “needle in a haystack” during model training, where it highlights fields that aren’t consistent with similar documents.
Use an ML-guided interface to train higher-quality ML models in less time and with fewer samples.
Learn More| Product Theme | Improvements |
|---|---|
| Ease of Use and Deployment | Performance and training experience improvements for multiple bounding boxes through guided data labeling and model finetuning. |
| Better visibility and flexibility of Flows: - View failed flows & automatic retry - For faster overall document processing time. - Public API - For importing and managing the lifecycle of Flows. - Import files of any type for custom blocks - Use our Flows SDK to create custom code blocks which reference external files (like a CSV file) that can be easily used in the Flows UI by a business user. |
|
| Superior Accuracy | Automatic QA Sampling - For higher machine confidence and better automation rates, reducing manual setup and accelerating time-to-value. |
| Korean language optimizations for best-in-class performance. | |
| New languages supported - Thai, Hebrew, Russian, Kazakh, Estonian, Czech, Slovak, Lithuanian, Latvian, Turkish, & Bulgarian. | |
| Enterprise Scale | Machine Learning Improvements - Including Faster transcription, proactive alerts for layout variations and improved signature detection and identification. |
| Reporting Improvements - A number of incremental reporting and data exporting enhancements, including the ability to bulk download submission activity logs from the submission table. | |
| New Integration with One Identify Safeguard - enables centralized secrets management in Hyperscience including encrypted storage, policy enforcement, role-based access control and audits. | |
| Human Centered Approach | Improved, faster decisions with required and dependent fields in Human-in-the-Loop tasks - Tackle more complex use cases with better handling of process variations and control over case routing. |
Full details in the release notes.