October 2024 - Enhanced AIDP projects, better data pipeline control and upgraded reporting.

This October release brings an improved AIDP project experience, better document processing, refined data pipeline oversight, time-saving AI features for document training and labelling, and enhanced reporting in Data Platform and pre-configured solutions.

Ingestion

Fine tune your parsing logic in AIDP projects

Parsing dates and numbers can sometimes be a challenge and depending on the use case it can be different for each type of document. Duco now allows you to configure how to parse these formats. In October release we’re introducing a smarter and configurable data parser, so you can customise it to better fit your needs. 

The new parser will always try to look at the context within a document to parse ambiguous dates and numbers. This means it will try to find non-ambiguous dates and numbers within a document to learn the format and apply that format to the ambiguous dates and numbers within that same document.

You can also enable the use of AI as a parsing fallback. This can be enabled when the parsing fails and/or when the parsing stops when dealing with ambiguous dates.

Feature available for Early Adopters. Please reach out to your customer success representative to learn more.

Spend less time annotating documents with new “annotation-less” models

Users who train their own models are able to enable models for common document types without any annotation. These “annotation-less” models use OpenAI to predict data entities in common document types like invoices, bank statements, and loan applications. This significantly speeds things up when creating your first model.

Feature available for Early Adopters. Please reach out to your customer success representative to learn more.

Document transformation improvements with processing rules

In this release we’re introducing more flexibility for processing data extracted from unstructured documents. You can now easily remove spaces, special characters or merge multiple occurrences of a single entity into one value. For example, if you have a billing address spanned across multiple lines or places in your document, you can now transpose it to a single value with just one click and spend less time doing data preparation further downstream.

Feature available for Early Adopters. Please reach out to your customer success representative to learn more.

Additional OCR engine

To bring you an even better document processing experience, we're thrilled to introduce Azure as a new OCR engine option. Azure OCR excels at recognizing handwritten text, forms where characters are filled in boxes, and general text recognition.

Azure OCR is one option, alongside Amazon Textract, that you can choose for the OCR portion of your projects. 

Feature available for Early Adopters. Please reach out to your customer success representative to learn more.

Improved processing orchestration in Data Prep

We’re introducing a new setting for time-based snapshot triggers. If at a snapshot time you still have unstructured documents pending user validation (AIDP submissions), you can now choose to not generate the snapshot automatically. This introduces more control to your data pipeline and could prevent you from sending unreviewed or incomplete data to downstream systems.

Reconciliation

Predictive labelling 

Early adopters, here’s your chance to transform your operations with Duco’s new predictive labelling feature. Our largest clients spend more than 5,000 hours on labelling alone — reviewing breaks, searching for labels, figuring out which label is the right one. Though each step takes seconds, the time adds up. Using this process, labels are often applied inconsistently, which complicates exception management and root cause analysis.

Duco’s AI-driven automation can reduce labelling time by up to 50% and increase consistency and accuracy, making investigations easier. It learns from your past activity, predicts the right labels, and customises suggestions to your processes. No searching or guesswork — just review, confirm, and apply in bulk, keeping humans in control.

If you wish to see how the new capabilities can enhance your exception management, please reach out to your customer success representative.

Data Access

Extract comments added to exceptions from Data Platform

Comments added to records and exceptions in our single and two sided processes can now be found in the Data Platform, allowing you to extract the data systematically for reporting, auditing and other purposes.

A new view records_comments is now available. More details can be found in the Data Platform documentation within the platform.

MAS and ASIC solutions

More insights into your reported data

Similar to the exception dashboards that we provide as part of EMIR solutions, you can now get insights to the reported data for the recently launched MAS and ASIC solutions too. They will now help you visualise your open exceptions from the latest runs of your processes built from these pre-configured solutions.

The dashboard includes exceptions views by

• Process
• Exceptions age
• Assigned group
• Workflow status

Drilling down, you can see details and access filtered results, enabling you to label, assign and take other actions quickly. You can also download what you see on the dashboard in PDF as a snapshot of the current exception state.

 

Fixed bugs