PIMSS Data Systems has launched a new module for its existing Compliance product which uses AI to extract data from documents to automatically update workflows and remedial works.
PIMSS Compliance AI Document Reader saves users time by manually extracting data from documents, removing the need for end-users to use templates for reading documents and run data-loads from spreadsheets.
Compliance AI Document Reader Using Microsoft’s Azure Cognitive Services to extract information from documents. The module recognises both typed and handwritten text, data spanning multiple lines and data flowing over separate pages. The system recognises where remedial actions are needed and creates work items within the system that can be assigned to internal teams and external contractors to be actioned.
The AI service is trained using the documents it needs to recognise by uploading only a few samples of a document and identifying the data needed to be extracted. When a document is sent for processing, it uses AI to identify the type of document and appropriate model to extract the data.
Automatic validation of each document associates the document with a specific workflow in PIMSS Compliance (via UPRN and associated workflow type). The document updates the workflow events in the system, including the creation of remedial works and actions, next inspection due dates and retaining the PDF document with the associated workflow.
Each line of data extracted from a document is given a confidence score for accuracy; any data falling below a certain threshold is automatically rejected and placed ‘on hold’ for an end-user to review. Documents which require user review can have any erroneous data corrected in place and then progressed to update the system.
