Document Management System with AI-Powered Handwriting Recognition

Client

The Client is a large drilling contractor operating a fleet of jackup rigs in the Gulf Cooperation Council.

Need for SharePoint Developers to Finish Building a DMS

The Client had document-intensive business processes that often involved external contractors. In order to control and standardize the document life cycle, the company hired a third-party IT vendor and started developing a SharePoint-based document management system (DMS). However, the Client had to part with the vendor halfway through the DMS development process and was looking for a new, competent SharePoint development team to finish the work. The Client found the required expertise at ScienceSoft and trusted us to resume the project.

Evolving a SharePoint DMS and Building a Document Control System

ScienceSoft assigned a team of 4 SharePoint developers (one of whom is a Lead developer), 2 SharePoint consultants, a project manager, and a UI designer to the project.

At the discovery stage, our team conducted an on-site visit to study the document management processes from the inside of the organization. This allowed us to collect the complete requirements at the initial stage of the project and make valuable DMS improvement suggestions.

Following the Client’s needs, ScienceSoft enhanced the DMS with the following capabilities:

Document transmittal creation workflow. The Client needed a formal, controllable way to exchange documents between employees and with external business associates. For this purpose, we enabled transmittals and built a step-by-step transmittal creation process. Users can select the documents for the transmission, fill out the necessary attributes, and attach a summary cover list to create a deliverable package.

Document version control and archiving process. The Client’s documents typically undergo numerous revisions and updates. To make all document versions uniquely identifiable, we implemented automated document numbering based on the Client’s internal version naming conventions. Also, to prevent confusion caused by multiple versions of the same document, the Client wanted to show only the latest document version in the document library. We built a workflow that automatically sends a document version to the archive once a new revision appears. Users have access to the archive with all previous document versions.

Optimized MDR workflow. The Client’s DMS utilized a master deliverable register (MDR), which was an .XLSX file with 16,000+ rows that contained the names and metadata of all external documents stored in the DMS. The Client used the MDR to create placeholders (empty documents with unique names) for new documents to be filled out by external vendors. These placeholders were later replaced by the completed documents when vendors sent them back. The Client used to have a workflow built on Power Automate that was too slow: it needed nearly 8 hours to scan 16,000+ rows in the MDR and create new file placeholders. We replaced the workflow with an alternative custom solution built on Aspose Library with Azure Function that executes the workflow in less than 10 minutes.

Generation of punch lists. We created a template for adding documents to punch lists. DMS users can select a document, populate the attribute fields in the template, generate a punch list, and send it for approval to a responsible party.

Document workflow improvements. We enabled the following functions:

• When a user uploads a document, document attachments are also uploaded automatically, and the document’s native files are added depending on the permission level.
• PDF versions are automatically generated for documents of different formats.
• Documents with a pending status are shown in a separate dashboard.
• Document attributes can be uploaded in bulk.
• Bulk downloading of documents based on the chosen attributes is enabled.
• All older revisions of a document are automatically labeled as superseded and stamped with a watermark.

Document Control System. Upon the Client’s request, ScienceSoft delivered a separate SharePoint site, named a Document Control System, which was needed to create document drafts before sending them to the primary document storage in the DMS. When a user creates a new document, the Power Automate flow automatically adds the update to the document change logs and assigns a unique name to the file. When ready, the document can be transferred to the primary DMS storage.

RFI Submission Portal With AI-Powered Data Extraction Engine

The Client needed a dedicated solution to receive and process requests for information (RFIs) from vendors without giving them direct access to the DMS. ScienceSoft’s team developed an external portal using Power Apps as a low-code, cost-effective platform that could be easily integrated with SharePoint. On the portal, vendors can submit RFIs via an input form and attach supporting PDF or .XLSX files.

When the Client shared its desire to automate the processing of RFIs, ScienceSoft offered to use AI for the task by leveraging Microsoft AI Builder. We used templates of RFIs provided by the Client to train an AI model to extract relevant data from the vendors’ RFI submissions, including handwritten notes, and save them in a unified format in a separate document. The AI model processes RFIs with 87–90% accuracy and facilitates significant time savings on manual data extraction.

We further streamlined RFI processing by adding a one-click approval workflow and automated notifications on RFI status for the Client’s employees.

Fully Functional DMS with Streamlined Workflows and AI-Powered Document Processing

After two years of ongoing collaboration with ScienceSoft, the Client has a robust SharePoint DMS for capturing, storing, and managing internal and external documents. Apart from finishing the DMS development project abandoned by the previous IT contractor, ScienceSoft introduced a number of improvements, including a controllable way of exchanging documents with external business associates, a 50x faster workflow for placing external documents into the DMS, and a tailored version control and archiving process.

The Client also received a dedicated RFI submission portal for its external vendors. To streamline the processing of RFIs, ScienceSoft introduced an AI algorithm for extracting the information and digitizing handwritten notes in RFIs, which demonstrates 87–90% accuracy.

Technologies and Tools

SharePoint Online, Microsoft Power Apps, Microsoft Power Automate, Microsoft AI Builder, Aspose Library, Azure Functions.