WorkPoint 365 makes it fast and easy to build modular solutions – with no development or coding.
WorkPoint 365 structures data via SharePoint and bridges to Microsoft 365 – As a result, you receive a seamless and future-proof cloud solution.
Together with the automated classification and inheritance of data, oversight and control is ensured, which makes it possible to scale WorkPoint 365 for an enterprise solution.
WorkPoint makes it fast and easy to archive and journalize mails. By integrating a drag and drop functionality combined with an intuitive search tool, this saves users in many hours of systemizing their mails.
Christian Vinter is the AI lead at NextAgenda, a consultancy firm that partners with companies and organizations to help them become more data-driven in their approach to competitive markets.
Part of printing and digital technology specialists and WorkPoint partner Konica Minolta since 2019, NextAgenda’s AI team is responsible for innovating print management and document handling.
With the world moving ever closer towards a paperless workplace, Christian and his team have been exploring the opportunity for helping businesses to stop printing, by creating analysis software that can categorize what’s been printed, by whom, and when. This provides some fascinating data insights and reports, which organizations can use to learn more about their user habits.
Building on this idea of understanding documents better, we saw the potential for expanding it out to become what is now known as ‘Document Intelligence’. But what is Document Intelligence exactly, and how can it benefit document handling?
Document Intelligence is an AI that analyses documents, understands what they’re about, and then assists with document handling. The AI doesn’t alter the content of the documents, it understands them, extracts information, and then takes action or enables a user to take further action. The Document Intelligence software is able to recognize different types of documents, including emails.
For example, one public sector use is in The Danish Confederation Of Trade Unions, which must be notified every time an employee is to be made redundant. Due to the reporting required, the organization receives enough emails to employ the equivalent of one person full-time to handle correspondence.
With training, Document Intelligence can recognize and categorize the nature of an email. It can then extract five data points: who is being made redundant, their position, the employer, the date, and their Danish personal number. It is then forwarded to the correct local union for further action.
Again, once trained, the AI can sift through documents like PDF files, determine their content, and extract pre-configured information. Created to recognize different document layouts and place them into a relevant category, the AI helps companies to organize their documents, and extract key data to accurately and efficiently archive them.
The incredible thing is that Document Intelligence was developed using existing open-source code. And for organizations who have not yet migrated to the cloud, it can be set up on an on-premises server, so that sensitive data does not go outside the organization, increasing its security.
Document Intelligence is currently helping to revolutionize document management and archiving in public sector organizations in Northern European countries like Denmark, Finland, and Norway. In the private sector, the same technology is being applied in different ways, such as filtering work orders into a CRM and ordering systems.
So what’s the vision for Document Intelligence going forward?
Firstly, it’s an incredibly efficient system. The AI can predict the correct category and extract the correct information for your given data points in at least 90% of documents. Any cases which can not be categorized are added to a manual workflow for human clarification.
Secondly, the universality of Document Intelligence means it can be utilized across multiple fields – any workplace or system where documents are handled. In fact, it can be applied in any situation where a document needs to be read, categorized, and scanned for key data points. This opens up exciting possibilities for future applications of Document Intelligence in document handling.
Although there are, of course, various types of organizations, many actually handle very similar kinds of documents. Customer orders are a prime example of this. When a customer sends in their order to a company, the information has to be manually entered into a CRM.
With Document Intelligence, the AI can extract the correct data points and automatically add them to the system with near-perfect accuracy. Imagine the potential of this when amplified to cover all your incoming emails.
Banks are another good use case. They have very standardized processes for sharing information, for example in the settlement of securities. But they still handle these documents manually, taking information from emails and entering it into their systems.
Insurance is also an industry that handles huge volumes of documents and data, and also uses standardized processes. Here, there is clearly enormous scope for the application of Document Intelligence to make document handling more efficient and cost-effective.
Governments and local authorities receive vast numbers of documents such as applications for buildings, work permits, visas, and so on. Imagine implementing AI which could identify the types of applications, categorize them, extract the relevant data, and then prepare them for a human decision-maker.
Utilizing machine learning also facilitates use within the legal system – for example in fraud cases. In this context, AI and machine learning are able to flag potentially fraudulent documents which may require human intervention. In this respect, AI is not diminishing the need for human workers, but making the work they do more meaningful.
How does it work? Whatever the application, the AI is trained to recognize specific document types and data points, so it’s able to handle very precise information. And it doesn’t miscategorize documents. If the received content falls outside the parameters set, it is sent for manual processing. Approximately, 90% of the categorizing and archiving is done by the AI, leaving around 10% of documents where the content does not align with one of the predetermined categories.
Whilst there are many early-adopters of Document Intelligence, uptake can sometimes be stemmed by a reluctance to move away from legacy document handling processes. On one hand, manual archiving is time-consuming, and there is significant room for human error. On the other hand, there is still sometimes mistrust around the use of AI to replace manual processing. This seems to be largely due to the fact that although AI models are trained to predict the category of a document, they’re not able to reason in the same way as a human.
For example, whilst the AI could identify a document as a driving license, it’s not able to explain why it has done so – and this can make some people uneasy. It also has a reputation for being over-complicated, when in fact once the AI is built and trained for your specific needs, it is relatively simple to maintain.
For a platform like WorkPoint 365 which helps organizations to manage their documents and processes, the potential for Document Intelligence is huge. It could help and support users through various processes, and ensure that information is correctly categorized and in the right place. In this way, AI supports part of the process rather than entirely removing the need for human involvement.
For example, in terms of HR functionality and job applications, Document Intelligence could identify that you’ve been emailed an application, extract the key data like their name, the job they’re applying for, and the name of the person hiring, ready for the next stage of the process. This means that part of the work is done for you, so you can focus on more important tasks – like deciding on the right candidate for the job.
All in all, the future implementation and integration of Document Intelligence in document management processes presents exciting possibilities for reducing errors, saving time, and improving cost-effectiveness.