Microsoft AI Explained

From the boardroom to the operating room, Microsoft is transforming the way we work with AI.

Photo by Andrea Piacquadio from Pexels

The rise of generative AI (GenAI) technology has led to a scramble among businesses everywhere as they seek innovative ways to utilize this groundbreaking technology within their operations. This isn’t the first time we have witnessed such disruption; technological advancement often brings about significant shifts in the business landscape, compelling organizations to adapt and innovate. As we have been talking with our clients we are finding that while many are intrigued by the vast potential of GenAI, they are struggling to understand where to begin, often feeling daunted by the intricacies and challenges.

Microsoft has been at the forefront of this new innovation sparked through their investment in OpenAI. It has brought to market two distinct service offerings:

Azure OpenAI — enterprise AI platform that allows you to build your own custom solutions leveraging GPT-4, GPT-3.5, ChatGPT, Llama 2, Codex, and DALL-E models within your secured Azure subscription

Copilot — AI assistant integrated into Microsoft’s software as a service (SaaS) offerings to help support people in their flow of work while respecting the security boundaries of what a user has access to within their role

With Copilot, Microsoft has significantly grown their overall AI portfolio by integrating GenAI capabilities throughout their entire product portfolio. Almost every Microsoft application from Word, PowerPoint, Excel, to Microsoft Teams now includes the ability to leverage natural language to create, summarize, and iterate on content created with the Microsoft suite of products and services.

Figure 1 — Microsoft’s current GenAI offerings as of July 2023

It can definitely be overwhelming to try to unpack all these product offerings and understand what they are, and even more importantly, understand the business benefit. Hopefully this article will help you better understand Microsoft’s current product offerings, potential use cases for the technology, and then what might be a next step to build a plan to take advantage of all these great capabilities.

Azure OpenAI

Azure OpenAI is Microsoft’s enterprise AI platform designed to meet the vast needs of its diverse customer base spanning from startups to Fortune 50 organizations. It includes a full AI Studio to help you plan, deploy, and manage your custom solutions all from within the Azure Portal.

Some of the highlights include:

• Security: there is no training of the enterprise models with customer data; the service fits within the existing security boundaries of your Azure subscription
• Scalability: leverages the elasticity of the Microsoft Cloud to help scale based on the needs of your business
• Supportability: fits within the standard management paradigm of your other Azure resources
• Capability: enterprise-ready access to all of OpenAI’s models including GPT-4, GPT-3.5, ChatGPT, Codex, and DALL-E as well as Meta’s Llama 2 open-source model

What we typically see as a path for clients to build their own AI solutions is as follows:

1. Start by building an understanding of the technology and its possibilities within the organization, including the differences between AI, machine learning (ML), and GenAI.
2. Begin focusing on internally facing use cases to help your business and technology teams appreciate the benefits and functionality of the technology in real, relevant situations.
3. Take lessons learned from your internally facing application and then begin working on your customer-facing solutions.
4. Evolve your current offerings by taking the learnings from those customer-facing solutions and continuing to innovate.

Figure 2 — Azure OpenAI use cases

Microsoft 365 Copilot

As someone who has been working in the portals and collaboration space for many years now, I am personally very excited for the release of Microsoft 365 Copilot. Microsoft 365 Copilot combines the power of large language models (LLMs) with data in the Microsoft Graph — your calendar, emails, chats, documents, meetings, and more — and the Microsoft 365 apps to turn your words into the most powerful productivity tool on the planet.
​
The key components of Microsoft 365 Copilot are:

• Large language models — recognize your text and predict the desired output (text, images, etc.)​
• Microsoft Graph — provides a pathway to your data including documents, meetings, email, etc.​
• Microsoft 365 apps — how you’ll interact with Copilot to create, summarize, etc.

Here are some really exciting use cases for Microsoft 365 Copilot:

1. Leveraging Copilot to create a statement of work in Microsoft Word based on a PowerPoint proposal, in your company’s standard format.
2. Engaging with Copilot within PowerPoint to create an engaging presentation, iterating on slide design and copy.
3. Asking Copilot to provide analysis around data within an Excel document to identify trends, or asking it to create a pivot table for you without you having to spend time Binging how to do that.
4. When returning from leave, you can ask Copilot to suggest the most important emails to respond to, and to ask it to block off time on your calendar to catch up on open action items.
5. As someone who is perhaps double-booked often, you can use Copilot to understand what key decisions were made during meetings you couldn’t make.
6. When finished brainstorming with a colleague, you can ask Copilot to extract the information from your Loop component and provide a concise summary that can be shared with your supervisor.

Hopefully these examples are starting to help paint a picture of how Microsoft 365 Copilot can help with content generations and summarization across multiple experiences to help deliver significant value faster while freeing up people to spend their time focusing on more creative work.

Figure 3 — Microsoft 365 Copilot features

Recently, at the Inspire partner conference, Microsoft announced list pricing for Microsoft 365 Copilot at $30/user per month. Analysts had been predicting pricing to be around $20/user per month, so this list price caused the stock to rise while they speculated it was too expensive. As someone who has been working in the Microsoft ecosystem for quite some time now, I am confident that this pricing is reflective of what Microsoft believes the market will bear, and also factors in what it anticipates the average revenue to be per client based on their various volume license discounts.

The business case for Microsoft 365 Copilot is actually quite simple. If you assume that the average person works somewhere between 1,880 hours a year (assuming 40-hour workweeks, and five weeks of vacation + holidays), then essentially the question becomes: Can you generate $1.53/day worth of value from Microsoft 365 Copilot? For some users in the organization, it will be very easy to prove value, especially for those who are content creators.

Examples of people in these roles may include:

• Financial analysts who actively create spreadsheets in Excel
• Corporate communications specialists who are trying to build messaging that’s easy to comprehend at various reading levels
• Marketing analysts or folks who need to build compelling PowerPoint presentations to help showcase new products, or services, or to help inform senior leadership
• Project managers and business Analysts who are actively working on requirements documentation, creating presentations, or performing analysis against multiple spreadsheets

For others—perhaps those who are content consumers—the GenAI capabilities might not be immediately compelling. Therefore, I believe we will see organizations taking a “production” pilot approach of purchasing a set of licenses to target across a couple of core functions to prove the value prior to making a larger enterprise purchase.

Sales Copilot (formerly Viva Sales)

Perhaps one of the most compelling products that was announced late last year under the former name of Viva Sales is Microsoft Sales Copilot. Dynamics 365 was released back in 2016 to compete with Salesforce, whose Sales and Service Cloud offerings have been Gartner Magic Quadrant leaders for many years now. While Microsoft has certainly gained some ground, Salesforce still remains the incumbent across many organizations. However, Microsoft 365 tends to be the incumbent from a document and collaboration perspective, making Sales Copilot even more compelling.

Sales Copilot is an add-in for both Microsoft Outlook and the Microsoft Teams client that helps you interact with your customer relationship management (CRM) data in either Dynamics 365 or Salesforce from those applications. The intent is to help reduce the amount of time sellers spend performing administrative tasks and shift that to focusing more on getting in front of customers and selling.

Sales Copilot features

Three of the key GenAI features included in Sales Copilot are:

1. From within the Outlook experience, you can click on an email and it will summarize the thread into key points of the conversation.
2. Sales Copilot performs an intelligent meeting recap, which summarizes key points, decisions, and next steps. It also uses AI to capture potential sentiment of the speakers and helps provide a listen-to-talk ratio to help coach sellers.
3. After hosting a meeting with Sales Copilot, you can leverage the intelligent meeting recap to generate an email to your contact and reference the key points highlighted in that meeting, and also provide calendar availability for when you might be able to reconnect with that buyer.

Additionally, some of the non-AI (but still quite powerful) features are the ability to share an adaptive card of an opportunity within a Teams chat so that you can, at a glance, provide the person you are messaging additional context. It also includes the ability to quickly add or update CRM contacts from Outlook, save email to CRM, and review opportunities tagged to a contact from within Outlook.

At this point in time, Microsoft has a very strategic advantage in the sales productivity space. Also, similar to Microsoft 365 Copilot, the people who would realize the most value from Sales Copilot are in a seller role. So this is a product that would be licensed for a subset of an organization rather than blanketed across all users.

GitHub Copilot

GitHub Copilot was one of the first products Microsoft brought to market to help developers accelerate their development with GenAI. GitHub Copilot integrates with the most popular integrated development environments (IDEs) like Microsoft Visual Studio and Visual Studio Code, JetBrains IDE, and Neovim. GitHub Copilot is powered by OpenAI’s Codex model, which was trained on a range of public code repositories. It draws upon its training to suggest code completions, helping to reduce the effort required in coding, especially for routine or common tasks. Its features include:

• AI-driven suggestions for code completion​
• Code snippet generation based on context and patterns​
• Automatically creates comments based on your code​ to assist with the software documentation process
• Real-time assistance while writing code​, such as providing recommended practices for performing certain operations
• No need to context-switch between your IDE and Stack Overflow, Google, Bing, etc., when writing code

GitHub Copilot is targeted at software and data engineers as part of their daily work. There are still many studies being conducted to help quantify the productivity gains, but some estimate between a 30% and 50% increase in productivity through the use of GitHub Copilot.

Copilots everywhere!

Microsoft’s strategy of integrating Copilot throughout their services is yielding some very compelling use cases across their portfolio including:

Copilot in SharePoint — SharePoint has been a cornerstone of Microsoft’s collaboration portfolio for decades now, and even for folks who have been using it for years, there are always new features available to help enhance the experience. Copilot within SharePoint allows you to explain the type of page you’re looking to create, and it will start to pre-populate with web parts and images, and even create some text for you to refine as needed.

Copilot in Viva Engage — Viva Engage (formerly known as Yammer) is great for helping connect people across an organization, and a very popular use case is helping to remove the barrier between leadership and employees. Copilot within Viva Engage can be used to help author posts to ensure that they’re readable, engaging, and a great starting point for those who suffer from writer’s block. It can pre-populate a post with starter text, which you can choose to further refine with Copilot or put the finishing touches on yourself before posting to an organization. It can also leverage documents as a starting point for the message — for example, earnings reports or press releases.

Copilot in Viva Goals — Viva Goals is Microsoft’s enterprise objectives and key results (OKRs) platform meant to help people understand how the work they’re performing is helping affect the greater organization. Copilot in Viva Goals is meant to help simplify the process of goal setting by providing recommendations for OKRs based on existing documents. Additionally, once these goals are created, you can use Copilot to help provide a status summary and suggest next steps to help drive a particular initiative forward.

Copilot in Viva Glint — Viva Glint focuses on helping leaders get a better understanding of employee satisfaction and sentiment through a series of research-backed surveys that can be orchestrated with your organization. Copilot within Viva Glint can help leaders summarize and analyze thousands of employee comments, and provide a conversational way to explore feedback by asking questions through natural language.

These are just a few of the products for which Microsoft has announced Copilot integrations, along with Dynamics 365, Power Platform, Power Automate, and Power BI. There’s even a Security Copilot being developed to help business leaders proactively and reactively respond to security threats within their organization.

Planning your pilot

When planning a pilot for a new technology, and especially with something as exciting yet potentially complicated as GenAI, it’s essential to follow a structured approach, measure potential business value, and identify what additional applications or opportunities are possible.

At Slalom, we’ve developed a structured approach for guiding our clients through an evaluation of GenAI capabilities to determine whether we can bring something to life that allows for additional business analysis for more refined use cases and opportunities to drive value.

Our approach starts with:

Obtaining executive sponsorship: Pilots are most successful when there are executives involved who want to sponsor the project. They are important stakeholders to manage, and can help unlock the additional financial and personnel resources or prioritization needed to enable a successful evaluation.

Aligning participants: In partnership with your executive stakeholders, identify individuals who will participate in the pilot. Ideally, you want to aim for a diverse group of business- and technology-oriented professionals to ensure feedback from multiple perspectives.

Level-set on Microsoft’s portfolio: Everyone is likely coming to this engagement with a different understanding of AI, large language models, ChatGPT, and everything in between. Therefore, it’s important to start with what the technology can deliver, how it fits within the grand scheme of AI, what are some of the design considerations and potential pitfalls, and what might be applicable business use cases to your organization.

Prioritize a business use case: This is perhaps the most exciting part, where you decide on the business problem you’re trying to solve. This could be anything from surfacing a chatbot that can return results from a simple database to taking a call center speech-to-text transcription and summarizing the main outcomes and overall customer satisfaction (a form of record enhancement never possible before at scale). The more complicated the use case, the longer it might take to build out the additional application functionality to produce a working pilot, so keep that in mind as you’re brainstorming. It’s also helpful to consider how you’re going to test the application to further inform what you decide to build.

Build or pilot your application: This next step will differ slightly based on whether you are going to pilot one of Microsoft’s many offerings like Sales Copilot, or if you choose to build your GenAI application.

1. Build: Leveraging your favorite flavor of Agile or iterative development, this step includes provisioning all the necessary supporting cloud resources to support your use case and the code needed to help drive the experience. Include users in testing to receive feedback early and often to ensure that the developers have a clear understanding of how users will interact with the application and what their expectations might be of the outputs.
2. Pilot: Plan your pilot, build your success criteria, identify your users, build your user personas, create your change management plan, and execute the plan, partnered with activating the pilot for your user population. Support them through the journey with office hours and a feedback loop, and continue to manage and monitor.

Document and share: Every organization has different expectations for pilot documentation, but typically a good readout includes:

1. Pilot objectives
2. Business use case overview
3. Architecture drawing of what was built
4. User feedback
5. Summary of measured business value
6. Pilot team’s recommendation for the next steps, including any investment requirements

Additionally, given the excitement around this new capability, you should consider making your demo available either live or through a recording so others can see what was built or deployed to help build excitement and solicit ideas for additional use cases.

Slalom can help

No matter where you are in your GenAI journey, Slalom can help you navigate this emerging landscape. We have been helping our clients with everything from educational workshops to hackathons, building out proof of concepts (POCs), and helping scale AI solutions for organizations. Learn more at slalom.com/ai.

Slalom is a global consulting firm that helps people and organizations dream bigger, move faster, and build better tomorrows for all. Learn more and reach out today.