days until DORA is in effect on January 17, 2025. Download your cheat sheet here.
ECI’s CIO Rich Itri and VP, Partnerships, Jamie Smith recently hosted a breakfast where they broke down everything our clients need to know to get ready for Microsoft’s game-changing AI solution, Microsoft 365 Copilot. Here are their tips for preparing your organization to make the most of the opportunity.
Generative AI may represent the biggest paradigm shift in how multiple industries operate in decades. For many businesses, Microsoft’s 365 Copilot – its package of AI solutions for the suite until recently known as Office 365 – will be their first chance to fully leverage the technology at scale.
As the scope of the opportunity comes into focus, we are often asked both how businesses can best get ready, and how companies can position themselves to harness the benefits quickly.
Here’s what you need to know.
In advance: Requirements for Copilot
Before we get into the implementation process, it’s important to know there are a few prerequisites you need to have in place.
Unsurprisingly, given Copilot is a set of Microsoft products, this translates in practice to requirements to already be integrated into their ecosystem, including:
In other words: you should think of Copilot as a new set of productivity tools that integrate into your company’s existing set of solutions, rather than a freestanding generative AI platform.
Step one: Organise
The key to activating Copilot quickly is in the preparation – it won’t be as simple as onboarding just another update or software generation.
The first stage is to get your team organised and to review your existing policies and procedures. Are they aligned with Copilot functionality, security and compliance considerations? What knowledge gaps do you need to fill – does your team have access to the information and training they need?
It’s sensible to appoint a steering committee or working group to oversee this and take a broad view of Copilot implementation, with the dual priorities of optimizing your operation for Copilot and ensuring you are in line with industry best practices. This group could also take responsibility for socializing industry-specific use cases and gathering information on these best practices.
This needn’t be solely an internal exercise: now is a good time to gather perspectives from your clients to establish clear expectations and requirements. What needs could you meet in new ways?
Step two: Assess
Next, conduct a thorough assessment of your current infrastructure, taking a clear view of where your current setup may fall short.
This is perhaps the most challenging stage, but is also the most critical: for example, Copilot will require the use of both Windows 11 and the new version of Outlook. Some integrations between Microsoft 365 and third party solutions may not survive the migrations as these third parties have not yet released updates compatible with Copilot.
Most developers are planning to offer this, but they will be available neither overnight nor all at the same time, so it’s important to identify which integrations may be lost, however temporarily, and to communicate these across your company.
Migrating platforms is not easy, with Outlook in particular a likely stumbling block – but it’s far better to identify these potential issues early on in the process so you can develop realistic timeframes and, if necessary, manage stakeholder expectations.
The other hurdle many firms may find challenging is the wholesale review of your metadata tagging. Despite excitable press coverage, large language models and AI aren’t magic, and you won’t be able leverage their Copilot instance to its full potential without maximizing the reach and quality of your tagging.
There are assessment tools on the market you use to evaluate this, alongside access controls and data loss prevention measures. Checking and if necessary updating permissions will also be a key piece of hygiene work – Copilot’s ability to search and surface data quickly and intelligently is redundant if the right people don’t have access to the right information.
Based on the results of these assessments, you’ll be able to create a strategy to remediate any issues in advance of deployment and solve many deployment issues before they arise.
Step three: Deploy
You should implement updated permissions based on your assessment’s findings, and rigorously test new access controls to ensure seamless operation.
Alongside data loss prevention measures to safeguard sensitive information, we recommend leveraging Microsoft Purview for metadata tagging, enhancing your data organisation and Copilot ‘s ability to retrieve it.
If necessary, now is the time to establish compliance reporting mechanisms; it will definitely be the time to provide comprehensive training materials and sessions so end users can familiarize themselves with Copilot.
After that, the only thing left to do is to enable Copilot across all your Microsoft 365 apps.
Step four: Manage
After deployment, it's crucial to actively manage and optimize Copilot's performance, but also to manage expectations.
The huge expectations around AI’s potential impact on productivity have meant that users may anticipate radical possibilities right ‘out of the box’. To get the most out of it, you’ll need to customize Copilot to the specificities of your data and what you want to do with it. Because it’s highly likely that curious and enthusiastic stakeholders will want to use it quickly, it’s important that you communicate that before Copilot is fully customized and optimized for your company’s needs, there may be a gap between its functionality and what they expect.
Otherwise, the final stage is very similar to what you would do after onboarding any other new software platform or solution.
You’ll need to keep a vigilant eye to make sure Copilot’s usage aligns with acceptable practices and to prevent any data loss incidents, while also fine-tuning policies and controls to adapt to changing business requirements as needed.
It is also advised to ensure you provide robust support to end users, addressing any queries or challenges that may arise, and being mindful that generative AI at an enterprise scale is still in its infancy. And, think about what new channels to greater productivity could open up that maybe you hadn’t anticipated before deployment?
