Artificial Intelligence for a Repository Neutral ECM


Artificial Intelligence for a Repository Neutral ECM

On a recent trip to California I discovered how convenient instant access to information from Google Assistant on my Pixel phone was to help make decisions in a place I was completely unfamiliar with. While navigating to the next stop I could ask for ‘places to eat’, ‘gas stations’ or ‘tourist stops’ and have suggestions, from data scattered all over the web, presented in real-time in Google Maps. Imagine if your ECM could do the same and present information and search results from all the different systems and repositories in your organisation in one simple familiar interface.

For this to work the ECM would need a common interface that connects to your CRM , Accounting System, shared network drives, file syncing services like Dropbox and OneDrive, e-mails and SharePoint as well as some way of reading all the content in those repositories and intelligently storing metadata to allow you to search on it. Combine the ability to add your own metadata to those items while preserving the content from its original repository so it doesn’t stop its use in the original system and you would have a very user friendly, ‘Repository Neutral ECM’ where the context is more important than where something is stored.

The figure on the right provides an overview of the ‘Repository Neutral ECM’ architecture that M-Files will release later this year with a vision that ‘Context is King’


The ‘Unified User Experience Layer’ is the ‘simple user interface’ that provides a single familiar user interface to interact with information regardless of the original repository. Think of it as Google Maps. This includes simple user access from any device including mobile apps for phones and tablets in addition to PCs. Just like Google Assistant’s ability to present outside information in Google Maps by simply asking, having a simple user interface means the user doesn’t need to learn other systems to be able to find relevant content in them and they can add their own metadata without stopping it from continuing to be used in the original system.

The ‘Multi—Repository Backend’ connects with the organisations repositories and systems via ‘connectors’ that include a set of core ‘out of the box’ connectors for repositories like network file shares, Office 365 and SharePoint, but also allow third-parties to develop connectors for other repositories and systems. This allows organisations to preserve legacy systems and avoid expensive integrations or migrations to new systems just to add functionality.

The ‘Intelligent Metadata Layer’ (IML) contains the intelligence components and multi-repository search along with the typical capabilities of an ECM such as search, dynamic views, workflow, security, version control and check-in/check-out. The intelligence components support automatic classification and metadata suggestions using text analytics. Like the Multi-Repository connectors, third-parties can add ‘metadata providers’ for specific industries or use cases. Along with text analytics, this layer includes machine learning to help improve suggestions based on user behaviour.

The power behind IML’s ‘Intelligence Components’ comes from the integration of Artificial Intelligence (AI) from Abbyy into M-Files. Abbyy produces Artificial Intelligence technologies based on textual content capture and OCR. This AI technology allows text to be understood and interpreted based on its content using algorithms that analyse the meaning of the words and the relationships between them. This allows accurate classification of complex and unstructured data in real time.


It’s exciting to see this automatic classification and metadata tagging in action, drag and drop a document into M-Files and you’re presented with ‘tags’ or ‘suggestions’ that you can click on to populate the metadata fields. Similar to how Google Assistant effortlessly presents pins on Google Maps of suggestions from your request on ‘places to eat on my route’.

The ‘tags’ are based on the content of the document being passed through the Intelligence Services in IML and returning matches. If you don’t like the suggestion you can still select metadata as you would in the past and the AI learns from your behaviour. This technology will improve the efficiency and accuracy of data typically entered by humans as the suggestions help you make the right selection.

The benefits of IML don’t stop at metadata suggestions, there’s also the External Connectors to other repositories. We’ve all used Windows folder search and most likely found it painful at the best of times, especially if it’s a network share. This is where IML’s External Connectors can help, because the content is indexed by the ‘Connector’ you can use M-Files powerful search feature to quickly locate a file based on its content rather than where you think it might be stored. It’s lightning fast and allows you to add your own metadata to any object from any repository to help you manage your information better. Having a connector for every repository in your organisation is a powerful concept that is difficult to ignore.


The Intelligent Metadata Layer allows organisations to have a true Repository Neutral ECM by providing Intelligent Services and External Connectors that present information from all the different systems and repositories in a single simple to use interface. It allows them to keep their legacy systems and avoid expensive integrations and migrations while providing simple efficient access. If you’d like to find out more on M-Files and how the Intelligent Metadata Layer can help your organisation, please contact us.

 Read my blog on 5 Things to Consider when Preparing for a Respository Neutral ECM.