Five Insights from Our Experience Using Microsoft 365 Copilot

By

on

In its rush to translate GenAI into reality, Microsoft has applied Copilot terminology to practically all its GenAI offering–from Azure to GitHub to Office. This blog focuses squarely on Microsoft 365 Copilot–an add-on to either the personal or business versions of Microsoft Office 365. Even in our early interactions over the past few months, it’s apparent that Copilot will eventually be a game changer, unlocking significant productivity and creativity. Reid Hoffman, co-founder of LinkedIn put it best, we should all be “experimenting vigorously and getting comfortable with how this new technology can change how we operate.”

Before we go too far, a quick recap on what Microsoft 365 Copilot is–it is an AI-powered digital assistant that integrates with all the Microsoft Office suite of applications. It is similar to ChatGPT and uses the same underlying engine; the big difference is that it can search across both your data (emails, calendar, chats, documents, etc.) and the internet. It is available as an add-on service with the Microsoft 365 packages, both for individuals and businesses.

1. Copilot for Teams is Already Redefining the Concept of Meetings

Copilot for Teams is the real gem of all the use cases and it’s hard to imagine what online meetings were like in the pre-Copilot era. We record most of our meetings by default now so that it automatically transcribes; if folks join the meeting late, they can ask Copilot to summarize the meeting so far to catch-up on what they missed. Post meeting, I routinely ask it to summarize key actions items or refresh my memory on the specifics of the discussion without pouring through the entire transcript.

2. Copilot Chat Redefines the Concept of Search

With a deluge of information that hits us across Teams messages, emails, and calendar attachments, it becomes hard to track data. I find myself using the web (www.office.com/chat) Copilot page as a great starting point to locate information quickly, track down any pending tasks (from my emails), and summarize my schedule. A rather common query for me before any client meeting is to ask Copilot to summarize my last interaction with the client. This is also a great platform to create and refine content.

3. Copilot for Word is a Great Scratch Pad for Idea Generation and Content Packaging

Whether in Outlook or Word, Copilot shows up by default when creating new content–an ever-present reminder that we don’t need to start from scratch anymore. Whether creating a job description or polishing the latest blog post on Copilot, it truly enhances personal productivity. In future versions, I can see how it would train itself on my voice by analyzing past content and applying that to new content–it’s not there yet.

4. Copilot for Excel and PowerPoint is a Work in Progress

While the overarching potential is that Copilot can analyze complex datasets to identify trends and generate reports, the reality falls short of expectations. To be clear, for simpler datasets, it’s much easier to create pivot tables, clean-up data, and query it. In our experience, it fell short on larger datasets with 1,000+ rows and 40+ columns. Microsoft acknowledges these limitations implicitly by indicating that Copilot is in “Preview” (aka work in progress) mode in Excel. PowerPoint is also limited in its abilities. It tends to overstuff the slides with text and is not innovative or creative when suggesting or creating images.

5. Instituting Strong Data Guardrails and Policies Becomes Paramount

Copilot applies user permissions to canvass and search across enterprise content as part of assembling its responses. If a user has inadvertent access to sensitive data (e.g., someone accidentally saved the payroll file in SharePoint without restricting permissions), it may surface much more readily as part of any relevant queries. Having strong data governance and an effective way to audit it becomes imperative in this new world of auto-discovery of data.

Other nits that we find annoying and hope that are addressed in future iterations:

  • Putting the Copilot activity in the background when it’s working effectively stops it.
  • Sometimes it’s faster to do the work on your own vs. wait for Copilot to respond–often in Excel.
  • Copilot still times out or gives errors and you must frequently tweak the prompt.

In conclusion, our exploration into Microsoft 365 Copilot has unveiled a glimpse into the future of productivity and collaboration. It’s evident that Copilot is poised to revolutionize the way we work, and is already in Teams meetings, chat interactions, and content creation within Word. While there are areas where Copilot shines brightly, such as its efficiency in handling simpler datasets and streamlining tasks, there are also areas where improvement is needed, notably in handling larger datasets and refining its suggestions in PowerPoint. Moving forward, it’s crucial to establish robust data governance to navigate the landscape of auto-discovered data. As we eagerly await future iterations, we remain optimistic about the transformative impact Copilot will have on our workflows and creativity, paving the way for a more efficient and innovative digital era.