You’re Ready for AI but is Your IA?

Ready or Not AI is here. How are you ensuring your Information Architecture is ready for Copilot?

AI (Artificial Intelligence) is here. Prompt engineering is the latest skill to start fine-tuning and Copilots are rolling out across Microsoft 365. It’s an exciting revolution in technology that may be the biggest change in how we work since the introduction of the computer.

But what does that mean for us? How does AI really affect the day-in-day-out of our lives right now?  

To be brutally honest, if we haven’t embraced the foundation of modern architecture and followed the guidance of Microsoft best practices for data management then we’re not in a place to fully realize the benefits of Copilot and AI for business and enterprise.

Information Architecture (IA) is not just how our SharePoint sites and Teams are structured. It’s how we organize, structure, and label our data.  

Over my career, I’ve trained and consulted on how to implement information architecture across various industries and organizations at varying levels of M365 maturity. Regardless of their differences, all organizations have these things in common: 

• Their people are busy 
• People don’t like change (it’s scary for many legitimate reasons) 
• Too much change at once overwhelms employees and can disrupt the business 

AI means change, but the real work is AI readiness. How many times have you heard that phrase in the last several months? I promise you that it is not just hype. If your information architecture hasn’t been prepared, you’re not prepared.  

What does it mean to be “AI ready”?  

Remember a time before AI and Microsoft Copilot became THE buzzwords?

Microsoft and consultants have been talking about flattening your site architecture, keeping folders fewer than 3 levels deep (I’ve been preaching that one for about 15 years), and organizing content into areas where the right people have the right levels of access through security, permissions, and sensitivity labels. 

That’s right… we’re really talking about Governance. Everything old is new again and AI readiness is another way of approaching governance with our content creators and business users.  

So, what are some of the things you need to do to get your environment ready for Microsoft Copilot and properly governed? Let’s do this!  

Make Sure M365 Licensing is Correct 

Trust me, I know how confusing M365 licensing can be. But it’s important to make sure the right people are properly licensed for the right tools. This process can help you in a few different ways: 

• Ensure employees have access to the right apps. 
  • This will aid in user adoption of technology and can help reduce shadow IT by ensuring they have access to the tools they need to do their job. 
• Verify that employees aren’t over-licensed. 
• It’s not uncommon to find out that some users have overlapping licensing. Cleaning this up can save you quite a bit of money and make room for the premium licensing that others need, such as SharePoint Premium and Copilot! 
  
  

Review Permissions 

Permission management is a common struggle for teams and site owners. Making sure content is shared correctly is a balancing act between too much and too little. Sharing links, granular permissions, and nested folders (often with granular permissions) are the bane of an owner and admin’s existence.  

There are new ways to handle content management and prevent oversharing in Microsoft 365. This blog highlights some of the capabilities now available. 

Things to review include, but aren’t limited to: 

• Areas with guests  

• • Should there be guests? 
  • Are those guests active/inactive? 
  • Do we know who the guests are and why they’re in the group? 

• Ownerless, memberless, and groupless workspaces  
• Too many group owners 

• • AKA “too many cooks in the kitchen” 

• Overuse of private channels 

• • Overusing private channels means content is locked behind unnecessary granular permissions
  • If files are being shared outside of the private channel, that’s a sign it shouldn’t be a private channel 

• Internal Sharing of files 

• • Review areas where there are granular permissions dure to internal sharing of content 

• • • Is there oversharing?  
    • External sharing? 

• Check default share link settings: ensure users aren’t oversharing with default settings 

• • For the organization 
  • For individual sites 

Permissions and security have always been important. Who remembers the first time a non-IT person saw Delve and lost their minds? It wasn’t showing anyone anything that didn’t already have permissions to, it’s just that user permissions were wrong, and no one knew until Delve exposed it. (RIP, Delve). I used to use that and some clever search queries to show the necessity of security/permissions reviews. 

With Copilot coming into play, permissions will once again need to be an area of focus. We must ensure that when Joy enters a prompt to roll up content from across the organization using specific key words, only the content she should see of the content she does see.  

The term “oversharing” is coming up often as Copilot makes its debut in our organizations. Typically oversharing of content happens when we use sharing links to share files outside the membership of a Team or site. Copilot will uncover and surface data using in-place permissions, even the sharing links. Knowing where these areas of granular permissions exist in your environment will be more important than ever.   

The other side of this is just as critical. If we have valuable information locked in private channels or group chats (instead of Team channels), Copilot will not help your employees harness it. 

Lifecycle Management 

In my role as the Director of Customer & Partner Success at Orchestry I have a front row seat when IT admins see the real number of Teams and SharePoint sites they have in their environment, and how many of those are inactive.  

Fun fact: The average number of active modern Teams and sites uncovered by Orchestry after the install is 33%. That means that on average 67% of workspaces are inactive. Is it just me, or is that a lot?  

When it comes to providing analytics, Copilot relies on a combination of techniques. These include but are not limited to: 

• Contextual Understanding: Copilot analyzes the conversation context, including the user’s query, previous messages, and any relevant information. This context helps determine the scope and focus of the analytics request. 
• Predefined Knowledge: Copilot has access to a vast amount of predefined information and tools. These internal tools allow it to retrieve relevant data based on specific queries.  
• Statistical Models: Copilot employs statistical models to predict likely responses. These models learn from patterns in the data and generate contextually appropriate answers. For analytics-related queries, Copilot might use statistical models to estimate trends or summarize data. 
• User Feedback: Copilot adapts based on user feedback. If a response is accurate or helpful, it reinforces that knowledge. Conversely, if a response is incorrect, Copilot learns from the feedback to improve future answers. 

Remember the old frustration with SharePoint search being less than ideal because of all the content it would bring back? Like that old Excel spreadsheet from 2004? How was SharePoint supposed to know you didn’t want that one?  

How will Copilot know what kind of content to not include in summaries? It will learn and refine results over time, but it will be up to us to ensure our data is well-manicured, properly permissioned and continue to validate responses over the course of its learning journey.

As we adopt AI for business, remember the following tip from the AI panel at M365 Community Days Miami. “Copilot is like having an intern or new hire. Follow up on its responses and double-check answers. Don’t count on it being right all the time.” 

What AI Readiness Really is... 

AI readiness is not just about having the latest technology. It’s about having a solid foundation of information architecture. This includes proper licensing, permissions, and lifecycle management.  

By following Microsoft's best practices for data management, organizations can fully realize the benefits of AI tools like Copilot.  

It is important to remember that AI brings change, but the real work is in preparing for it. By embracing governance and modern architecture, organizations can ensure that they are ready for the exciting revolution in technology that AI brings. 

Orchestry Helps Businesses Get AI Ready

Orchestry already provides the ability to facilitate governed workspace provisioning, the creation of archival/renewal policies, and secure guest management. We’re about to take our reporting and insights to the next level.  

Orchestry Recommendations is a comprehensive data collection engine to help organizations not just prepare for AI and Copilot, but to help them achieve a solid foundation in modern information architecture.  

Recommendations will take our reporting capabilities further and provide you with the steps you need to take to resolve issues such as: 

• Workspaces with inactive guests 
• Ownerless/memberless workspaces 
• Inactive workspaces 
• Workspaces near storage limit 
• Shared links reporting 
• .. and more. In fact, 35 more - We are releasing 40 recommendations in Q1 of 2024, with more to follow throughout the year!

AI for Business & Enterprise: The Future is Bright, But There’s Still Work to Do 

Microsoft Copilot and other AI for enterprise and business can be extremely useful for us in daily tasks, analytics, and other things, but we should not forget that we, the humans of IT and the business, are still the ones in charge. We have the responsibility to ensure the data that AI returns is correct. The best place to start is with tried-and-true, best-practice organizational governance. Otherwise, we are not as AI ready as we think.