7 Tips to Get Your Business Ready for Project Cortex

Microsoft’s Project Cortex is the latest competitive advantage in the workplace. Using artificial intelligence, Project Cortex captures and automatically organizes content from across the organization into topics, which are then delivered through topic cards, topic pages, and knowledge centers—making it easy for people to access valuable insights that are usually hidden away in documents, meetings, and conversations. 

However, to fully leverage a new and improved knowledge network that connects the right content with the right people, you’ll first need to review, rethink and update the information you have in your business systems to ensure Project Cortex only works with current, reliable data.

To help you get started on the right foot, here’s what you can do to prepare your organization before and on the first day of launching Project Cortex.


Getting ready for Project Cortex

As with any major addition to the workplace, integrating Project Cortex means refreshing your current setup. To avoid long delays and interrupting your organization’s workflow on launch day, you can start with the following:


1. Migrate relevant content

Project Cortex promises to empower teams with the right insights at precisely the right time, but it’ll only be as helpful as the content you provide. So, your first task is to make sure the information your team needs is already in Microsoft 365. This could mean migrating content from other systems (if it needs to reside on-premise, content can be brought in through search connectors), cleaning out or archiving old files, updating company terminology, and finally deleting the profiles of people who left the company months ago. 


2. Modernize your SharePoint sites

Many of the features in Project Cortex are based on the modern SharePoint UI, which was redesigned to be faster and easier to use, as well as responsive on mobile. If you haven’t already, you should modernize your classic SharePoint sites to leverage Project Cortex’s AI-powered content capture and categorization.

It’s advisable to start with the most frequently used sites. If you’re not a developer, you can get ahead by planning what modern pages need to be created. 


3. Identify roles and permissions

As an organization grows, so does the list of users and their roles. Now is a good time to ensure the right people have the permissions, responsibilities, and training they need to handle what Project Cortex has to offer. 

There are three main roles you need to consider:

  • Content Services Admin: This is the person responsible for assigning access and managing settings for content across M365. Usually, it’s the same person as your SharePoint admin.
  • Subject Matter Expert (SME): This is the person in charge of training AI models to teach Cortex what information to extract and how best to process it. They also help with curating content, such as topic pages that relate to their expertise.
  • Knowledge Manager: This person safeguards the quality of your content by managing topics, excluding sensitive information and ensuring all your content is accurate and up-to-date. In addition, they also champion awareness and adoption of new features to set your team on the right path.

4. Define your taxonomy

As part of its Knowledge Management service, Project Cortex builds on your taxonomy and managed metadata—so the cleaner they are, the better the AI will perform. If you already have a taxonomy, make sure you understand what terms you’re using and whether they accurately represent your range of content. If you don’t have a taxonomy, Cortex will create its own that can then be curated to define relevant topics. 

Keep in mind that a well-designed taxonomy should not only be logical and intuitive to use, but also go hand in hand with governance documentation and planning. Without proper governance, you could end up with inconsistent indexing/tagging, which means people won’t easily find the content they’re searching for. 


5. Improve your information architecture (IA)

Site maps, hierarchies, categorizations and navigation are just some of the elements of a well-planned information architecture. It’s the foundation of an efficient knowledge base and drives productivity by helping people perform their tasks with minimal effort.

Typically, creating your IA is a painfully long task that involves hours of brainstorming with people from different departments, battling with confusing software and importing/exporting spreadsheets. Alternatively, you can enlist a user-friendly tool like THEMIS IA, which makes it easy to import your existing IA, create one from scratch or build one using their intuitive blueprints. It even has a chatbot to guide you every step of the way. 

You can make changes to your design in just a few clicks and, once you’re happy, quickly deploy your new IA to SharePoint (or any other CMS).


Getting started with Project Cortex

On day one of Project Cortex, there are a few things you’ll need to set up to begin connecting the right people with the right insights. Namely, the two centers that Project Cortex populates after extracting important information from across your Microsoft 365: the Knowledge Center and the Content Center.

Depiction of the two centers created by Project Cortex.

In brief, the Knowledge Center is the central “wiki” where all your organization’s content is intelligently tagged (based on your own taxonomy, if you have one) and then categorized into topics, projects and their related resources (people, documents, media, chats, etc.). The Content Center is where your assigned SMEs can teach the AI how to process certain content types, so it “knows” what to do with them next time.

6. Configuring the Knowledge Center

One of the ways Cortex intuitively delivers content to your team is through topic cards, which pop up with key information when you hover over a new term in a document or email, and topic pages, where you can find all the people/resources associated with a topic. 

Cortex automatically creates these topics through Knowledge Mining, so the first thing you’ll want to do is define what content you want it to seed topics from. Then, you can go down this list to cover all your bases:

  • Develop a well-defined taxonomy to improve knowledge mining of topics
  • Exclude sensitive content from being mined
  • Assign users responsible for creating, editing, and managing topics
  • Define who can see topics (e.g. Everyone, specific groups, etc.)
  • Create a blacklist of topics you don’t want to be shown. 

In addition, you can include content from outside Microsoft 365 using Microsoft Graph Connectors.

7. Configuring the Content Center

Content Centers are SharePoint sites where you can train machine learning models to automatically extract and classify your content at scale—but it’ll need some human help to get started. Here’s what you need to do:

  • Curate content types and metadata

Machine learning in Cortex is driven by SharePoint Syntex, which uses AI to automatically recognize content types, extract important information from them and apply metadata tags. These tags are then used to organize your content and highlight important terms anywhere in Microsoft 365. 

One item on your organization’s to-do list should be to curate these AI-generated tags using the SharePoint Term Store, as well as create and manage your content types in the Modern Content Type Gallery. Both of these admin features are available in the modern SharePoint UI, which you’ll already have since you wisely updated it beforehand.

  • Build AI training models

To train the AI how to process specific content types, you need to build AI models for them. With SharePoint Syntex, your SMEs can easily build these models without writing a single line of code. They can also accelerate the process with the help of these AI engines:

  • Document understanding: Reads unstructured documents, such as manuals, contracts, or resumes, and learns what information to extract. Document understanding models can be published from the Content Center and used by many SharePoint sites.
  • Form processing: Reads structured documents such as invoices, processing forms and purchase orders to lift specific values (e.g. dates, names, figures). A form processing model can only be used in the document library it was created in.

If you’re not an SME, you can also make headway by planning what content types will need sensitivity labels and retention labels for automated security governance. (This feature is set to be released later this year.)

  • Streamline content management with specialized tools

Even with all these new and exciting features, building AI models and curating metadata can still take a long time and pull your team away from more pressing tasks. To speed things along, you can enlist a tool like THEMIS ICE to intelligently tag all your new and existing content, and then automatically migrate it into the right places within SharePoint.

Furthermore, THEMIS ICE accelerates the modeling development process by using pre-built datasets to help Cortex learn much more efficiently. With done-for-you AI models and practically hands-free knowledge management, THEMIS ICE is a worthwhile addition that will extend the power of Project Cortex to the benefit of your entire organization.


Dive Deeper Into Preparing Your Organization For Project Cortex

Checking off the items above will certainly give you a head start, but there is still more to be done before, during, and after welcoming Project Cortex into your organization. 

For a complete guide, myself (Cham Tanh, CTO) and Creospark VP Technology & Co-founder, Noorez Khamis, recently hosted a webinar covering all of this and more. A recording of the session can be found here, and I encourage you to check it out and learn how to deploy Project Cortex without delays so you can unlock organizational knowledge and empower your team as soon as possible.

HELUX would be more than happy to support your organization along its Digital Transformation Journey, get in touch today!