Copilot for Microsoft 365 (Copilot M365) provides AI capabilities within the Office suite of applications: Teams, Word, Outlook, Excel, PowerPoint, OneNote, Forms, Loop. But this comes at a cost of $30/user/month, and an annual commitment is required.
The key question is: What do you get for each user’s annual $360 expense?
I’m certain that no one ever accused Microsoft of making their licensing straightforward.
Copilot licensing can be especially confusing because there are many different Copilot licenses, and some capabilities Microsoft refers to as Copilot that are distinct from licensing.
Firstly, there is a free version of Copilot. It is free and honestly, it is pretty great. The free version of Microsoft Copilot brings ChatGPT large language models (LLMs) to help with searches, creating content, generating images (using DALL-E 3), and summarizing or asking questions related to web pages or documents. Copilot works on your PC, mobile phone, and in a browser. The free version of Copilot does not include integrations within the Microsoft Office apps you use.
Second, there is Copilot Pro which is designed for individuals or those with a Microsoft Office for Family Plan. Copilot Pro, priced at $20/user/month. Users can subscribe monthly, an option not available to Copilot for Microsoft 365 users.
Copilot Pro promises faster responses during peak times, faster AI image generation, and integration with some specific Microsoft 365 apps (Word, Excel, PowerPoint, Outlook, OneNote), but only when using the web versions of the apps. If you have a separate Microsoft 365 Personal or Family license, then Copilot Pro integrates into the same list of desktop apps.
Note: Excel Copilot features are in preview and only available in English. Therefore, it’s hard to know if this is worth $20/user/month.
Next, there is Copilot for Microsoft 365.
For most medium and larger organizations who already use Microsoft Office, Copilot for Microsoft 365 (Copilot M365) is likely to be the Copilot license being considered. As noted above, this license is priced at an additional $30/user/month and requires an annual commitment.
In contrast, Zoom does not charge additional for its AI Companion. Similarly, Cisco does not charge additional for its Webex AI Assistant.
Copilot M365 adds AI features to the Office applications Teams, Word, Excel, PowerPoint, OneNote, Outlook, Loop, and Forms.
The key and most powerful integration is between Teams and Copilot M365. Within Teams, Copilot can summarize 30 days of chat content, help compose or rewrite chat messages, summarize meetings, help you prepare for upcoming meetings leveraging the Microsoft Graph (details below), summarize calls, and help generate ideas when using the digital whiteboard.
Context is what helps any LLM deliver relevant results. “Grounding” is the specific term used to describe providing a LLM with information that is use-case specific, and not available as part of the LLM’s pre-trained knowledge. The Microsoft Graph provides a single point of access to the data stored in Microsoft Office applications, and interactions you have had with different people inside and external to your organization.
Because Copilot M365 is grounded using the Microsoft Graph (and optionally the web), Copilot can return more accurate information. For instance, if you ask Copilot M365 to help you prepare for your upcoming meeting with John, data from the Microsoft Graph helps identify the specific John you are likely referring to and can also provide information from previous chats, meetings, and emails, along with documents that may be relevant.
On March 22, 2021, Microsoft published its first Work Trend Index. This report provided insights into the evolving nature of work, especially in the context of the shift to remote work during the pandemic. Ever since, Microsoft, via its WorkLab site, has been publishing survey and research related to changes in the workplace. Not surprisingly, AI is a topic of focus.
In an April 2023 Work Trend Index Special Report, Microsoft recommended organizations “Use next-generation AI and data-driven analytics to increase the effectiveness of communications.” Although this recommendation was not supported by specific survey data.
In a May 2023 Work Trend Index Annual Report focused on the question “Will AI Fix Work?” The survey data from 31,000 workers across 31 markets showed a desire for AI to help deal with workloads. Some of the results include:
Overall, 49% of people worried AI might replace their jobs, but most, 70%, would still delegate as much work as possible to AI to reduce their own workload.
These desires and aspirations captured by the survey however do not prove that Copilot, or any other AI assistant, does reduce administrative tasks, help with analysis, or assist in creativity.
By November 2023, eight months had passed since Copilot M365 was introduced and Microsoft published a new Work Trend Index Special Report described as “A first look at the impact [of AI] on productivity, creativity, and time.” Of note, this report only included responses from 297 customer respondents who were part of the Copilot Early Access Program.
The results seem positive. Two-thirds of respondents (67%) said Copilot saved them time so they could focus on more important work, the typical user reporting saving 14 minutes a day, or 1.2 hours a week. This time was re-invested by most (53%) in focus work, with 16% of respondents using the extra time for additional meetings (perhaps indicating any AI-generated time savings will need to be properly directed).
Microsoft then created several simulated exercises to better quantify savings with Copilot M365, for each approximately 200 participants were involved. Copilot users were overall 29% faster completing an exercise that involved searching for information, summarizing a meeting recording, and writing a blog post. Most notably, Copilot users were able to summarize a meeting they missed almost 4x faster. Emails written with the assistance of Copilot were rated as 19% more concise and 18% clearer than those written without.
The latest Work Trend Index Annual Report, “AI at Work Is Here. Now comes the Hard Part”, published in May 2024, again focused on AI and carried the sub-heading “Employees want AI, leaders are looking for a path forward”, accurately reflecting the desire for AI along with the uncertainty in terms of how to proceed, and how to justify the required investment.
This latest Work Trend Index is again based on 31,000 workers across 31 markets.
The report highlights that 75% of knowledge workers are already using AI at work, with 78% of these users bringing their own AI tools (BYOAI). The subtext from the report being that if you don’t deploy a corporate AI tool, “shadow AI” will emerge.
Users of AI say (justify?) that AI helps them save time (90%), focus on important work (85%), be more creative (84%), and enjoy work more (83%).
Despite the qualitative opinions, most leaders (59%) are uncertain as to how they can quantify productivity gains from AI. This is likely why the survey reports almost half (45%) of US executives are not currently investing in employee AI tools, only 39% of global workers who use AI have received company-provided training, and only one-quarter of companies plan to offer generative AI training this year.
The quantitative signals remain unclear. As part of this latest Work Trend report, Microsoft conducted a six-month randomized study of 60 Copilot customers across multiple industries. Results were mixed:
A July 29, 2024 post from Judson Althoff, Microsoft Executive VP, provides additional quantified AI case study data; however, this article comingles data from self-created AI models (using the Azure OpenAI service) and usage of Copilot M365 (pre-trained) models. Extracting the key relevant examples (both quantitative and qualitative):
Microsoft understands that some customers may have concerns related to copyright infringement when using generative AI models such as Copilot.
In a September 2023 blog, authored by Brad Smith, Vice Chair and President and Hossein Nowbar, CVP and Chief Legal Officer, Microsoft stated “We are charging our commercial customers for our Copilots, and if their use creates legal issues, we should make this our problem rather than our customers’ problem.” Brad Smith, Vice Chair and President, Hossein Nowbar, CVP and Chief Legal Officer.
The blog went on to say, “Specifically, if a third party sues a commercial customer for copyright infringement for using Microsoft’s Copilots or the output they generate, we will defend the customer and pay the amount of any adverse judgments or settlements that result from the lawsuit, as long as the customer used the guardrails and content filters we have built into our products.”
It is not clear if Zoom or Cisco offer this type of copyright indemnification when using their generative AI tools. Google in October 2023 announced it will indemnify users of its generative AI products if they’re hit with claims of intellectual property infringement. But then, Google also charges $30/user/month for Gemini AI enterprise users, or $20/user/month for business users.
Does Copilot M365 deliver value that exceeds its cost?
It depends. At present the data does not support purchasing Copilot M365 licenses for everyone in most organizations based on productivity improvements. In these early days, there is lots of anecdotal data, a fair amount of qualitative data, and a very limited amount of quantitative data.
Based on self-reporting, some users note saving a material amount of time, improving creativity, and/or enjoying work more using Copilot. The challenge is correctly identifying and supporting these users.
Tying paid licensing to copyright indemnification is an interesting add-on benefit Microsoft is claiming but it is difficult to assess the value of this.
This is not to suggest that Copilot for M365 or other AI assistants will not be transformative for many workers; but rather, to acknowledge that we collectively have much to learn before cost-effectively leveraging AI assistants.
Experimentation, training, and analytics combined with a business focus and senior leadership are certain to be key elements in organizations that succeed in delivering a positive return on their AI investment.
Editor’s note: Check out this article on real-world generative AI applications and this piece on how Gen AI can increase productivity.
