12.10.2023

Copilot: Are you ready for take-off?

AI is one of the trending topics, including Microsoft Copilot. How can companies benefit from Microsoft Copilot and what are the key considerations?

If there’s one thing that has really generated a buzz this year, it’s artificial intelligence (AI). Starting in December of last year, when ChatGPT burst onto the scene, hardly a week has gone by without there being some new headline or challenge having to do with AI. Everyone was thrilled by ChatGPT, with many even calling it the dawn of a new era. But it wasn’t long before the challenges of large language models (LLMs) became clear, too. Racism, assumed gender roles, and misinformation were especially common – the issue there being that LLMs learn from existing data. If the data is erroneous or bad, then the output isn’t as desired, either. These issues were quickly and optimally resolved and are being continuously improved through intensive reviews, revisions and re-training of the models, including content moderation.

And then came the next big thing! Microsoft Copilot here, Microsoft Pilot there. Suddenly, there were a string of Microsoft Copilot announcements: for the Office suite, in CRM and SharePoint, for security topics, for Windows – the list goes on and on.

In a nutshell, Copilot will be everywhere, essentially acting as your right hand for your day-to-day work. I am certain it will revolutionize how we work forever. 

  • Late for a meeting? No problem! Copilot gives you a summary of what has been discussed so far.
  • Inbox bursting at the seams after your vacation? No problem! Microsoft Copilot scans your inbox and gives you an overview of the information you missed.
  • Writer’s block again? No problem! Microsoft Copilot provides text suggestions based on other contextual information such as OneNote notes or other documents. 
  • And many more productivity and creativity boosters.

 

Unlike ChatGPT, Copilot can extract information from your company’s data and enrich it with the current context of existing and available company information, without releasing the data or information to any external entity. That’s different from ChatGPT, where the LLM continues to learn and improve through user input, something known in the industry as reinforcement learning. The good news for companies is that Copilot can ensure that security and compliance guidelines are observed. But there are still challenges in dealing with misinformation and the correct use of the tools.

Just like before a flight: a checklist and preparation are crucial

Would you feel safe if you were on a plane and knew the pilot and copilot hadn’t checked everything before takeoff? Neither would I! The same is true of your personal copilot. Preparation is the key to success. After all, we learned one thing from ChatGPT: “garbage in, garbage out” – or, even worse: with the right prompting, you can get your hands on information not intended for you at all.

What is prompting?

What that means for companies is that good planning is essential when it comes to using Copilot. The license costs are considerable, at $30 per user per month, so a clear approach is especially crucial.

We recommend conscious use and a specific strategy for these tools. This can be done directly through resources at a company, or experts like those at Campana & Schott can be brought in. We follow a standardized approach, much like a pre-flight check. However, this check must be performed not only from the technical perspective, but also organizationally.

 

The organizational checks revolve around strategy, use cases, and business value analysis. Why does the company want to use Microsoft Copilot, for which use cases does it make sense, and will this in fact bring added value for the organization? All of these questions lie at the heart of readiness, and all of them also rely on the active involvement of the relevant decision makers so they can be brought along on the journey.

 

In our standardized approach, the technical checks are aimed first and foremost at making certain the files are available via SharePoint, the intended rights construct has been implemented correctly, and service and maintenance during operation is ensured through suitable governance structures. The key idea here is that of the “data lifecycle concept,“ which will be crucial when it comes to high-quality, sustainable use. This concept needs to be reviewed or, if not yet in place, introduced.
 

Readiness check passed. What now?

We have learned from many introductions of Microsoft Teams that the mere fact that a tool has been launched in technical terms does not mean it will be used properly. Far from it, in fact. The situation with Copilot is similar. Clear adoption and change management support is the key to success here.

One of the key factors is teaching employees how to work with the AI tools as they go about their daily jobs. How does the right kind of prompting work, how do you gain the skill to be able to gauge answers correctly, and when does it make sense to use Copilot – or not? In Copilot Labs, Microsoft provides a prompt repository with that can be used as one of many building blocks for prompt learning.

There is a risk of adding another communication channel that, in the medium term, creates confusion and uncertainty outweighing the value it adds. 

If the use of Copilot is planned well and appropriate measures are taken to introduce it, it can improve any organization, making it more efficient and revolutionizing daily life even in the short term. All of this comes in the service of a single goal: to support employees in everyday situations, freeing up more time for them to concentrate on the essentials.

Thanks to the standardized Campana & Schott approach, we ensure that the “flight” gets people from point A to B as efficiently and sustainably as possible, and that the journey is a pleasant experience as well.

Author

Pascal-Brunner_500x500.jpg

Pascal Brunner

Head of Modern Work Switzerland

Additional links