Enhancing efficiency and collaboration with Copilot
A case study discussing the department-wide trial of Copilot in IT Services
Key points
- Since the IT Services pilot began, Copilot has been saving time for staff across the department and continues to do so.
- Staff training and onboarding were crucial components of this trial.
- Providing resources and opportunities to engage and try new things encourages uptake as well as improving confidence and understanding.
- Not all Copilot users progress at the same rate, they use the tool in different ways.
- Copilot is evolving every day, what doesn’t work as expected one week, might work well the next time you try it. Learning how to ask the right questions (prompts) is a valuable skill.
As part of Oxford University’s digital transformation, generative AI is being explored across many university operations, from teaching and research to administration and infrastructure.
This voyage of discovery has included a university-wide generative AI pilot which included IT Services staff participating in a department-wide trial of Microsoft's Copilot, an AI-powered tool designed to streamline tasks and enhance productivity. The trial is exploring the potential and assessing the impact of Copilot in various aspects of IT operations, from routine task automation to advanced data analysis and everything in between but most importantly looking at how an entire department can embrace the opportunities offered by this.
Facilitating access to Copilot
To get underway, IT Services purchased 250 licences from its budget to facilitate a widespread trial of Copilot across the department. Allowing as many staff as possible to have access was an important part of exploring Copilot’s functionality and demand for licenses grew as the trial progressed.
Most importantly, the trial had senior level support, with regular communications from the Head of Department encouraging use and keeping colleagues informed and engaged. Prof. Stuart Lee, Interim Chief Information Officer, explained:
When first trialling Copilot, members of the department approached it with a mix of curiosity and apprehension. Many of staff members were eager to get involved, while others were uncertain about how AI could be used and what it might mean in terms of efficiency and improving workload pressures.
Encouraging exploration
Encouraging discovery and exploration was a key objective, but there was clearly work to be done to reassure and encourage staff to take an active role in the trial. To help with this, a Copilot Adoption Team (CAT) was set up to help colleagues, answer questions, demonstrate successes, and promote Copilot within IT Services.
Led by the Digital Innovation team, CAT members encouraged an experimental and collaborative mindset, aiming to build confidence among Copilot users in the department. They wanted license holders to engage with Copilot, explore the possibilities and become comfortable with using this new tool.
Understanding how to get the most out of Copilot for practical day-to-day tasks has been a key strand of the trial, but equally important was discovering what didn’t work so well.
Providing training and building confidence
To help colleagues across the department build confidence, the Copilot Adoption Team encouraged community engagement and delivered a series of comprehensive training activities in various formats.
Early in the trial, a department-wide Copilot Teams channel was set up where colleagues could share tips and ask questions. Regular features included a ‘prompt of the week’ competition encouraging colleagues to share their best prompts, and a ‘cartoon of the week’ intended to inject a little humour and inspire use of the image-generation feature of Copilot.
The CAT members curated an extensive prompts pool to help colleagues understand the potential of Copilot. They also arranged onboarding meetings, briefing sessions and practical demonstrations showing how to use Copilot with regularly used apps such as Word, Excel, PowerPoint, and Outlook, providing practical, step-by-step guidance to help build user confidence. There were even sessions on how to build bots using Copilot, helping enhance technical skills, sparking creativity, and resulting in the development of prototype bots to address specific departmental needs. The department continues to explore and learn about embedding Copilot agents in everyday workflows.
Drop-in clinics were available every week to help colleagues work through particular challenges or simply introduce them to the basics of Copilot. The team also created personas intended to inspire and demonstrate how Copilot is used in typical scenarios or for certain roles within the department.
These promotional and practical activities were supported by a series of podcasts by colleagues sharing their real-world experiences of incorporating Copilot into their everyday work, as well as internal articles highlighting Copilot's capabilities, benefits and limitations. These included links to external resources and deep dives into specific functionalities.
It was important to demystify initial complexities and inspire users to try new things. Focusing on practical use cases, targeted training, and effective communication, helped to build a strong foundation for broader Copilot adoption across the department and those participating started to experiment more with Copilot as part of their day-to-day work.
Regular monitoring for continuous improvement
Regularly monitoring Copilot’s performance and gathering user feedback was important, particularly as Copilot is an evolving tool that changed considerably during this trial and continues to do so.
During the trial, engagement levels were monitored and the team worked closely with the AI and Machine Learning Competency Centre, which provided valuable insights into user expectations and adoption.
In December 2024 a week-long survey was conducted, asking IT Services licence holders to track their use of Copilot and the time saved on a range of activities such as summarising content, generating material, managing meetings, or handling tasks.
Responses from engaged users showed the potential time saving benefits of this exciting new tool, especially when it’s used regularly. The results from 61 respondents suggested that they could be saving up to two hours per user per week. Microsoft themselves suggest that on average they see in other units that time saved is about 14 minutes a day so this suggests with proper support this can be increased.
A survey will be run at the end of the pilot and the benefits of Copilot within the department will continue to be monitored.
Pros and cons
One team member explained how they often use Copilot to create a rough first draft of text, compiled from multiple sources. She explained:
It’s only as good as the prompt I write (I’m getting better at them!) and often I end up rewriting the text. But it provides a quick starting point, much faster than when I do it myself, especially if I know the topic well and can quickly spot errors or omissions,
Other respondents found Copilot useful for various everyday tasks, such as locating lost emails, conducting internet and file searches, writing code, developing lists of questions and answers, summarising lengthy documents or meeting notes, generating questionnaires or first drafts of standard documents (job descriptions, terms of reference), and planning coaching sessions for their teams.
But while Copilot offers many benefits, there are some areas where it doesn't work as well. For instance, users have reported issues with Copilot's integration in certain applications like Excel and Planner. In Excel, Copilot sometimes struggles with complex data tasks and may not provide accurate or helpful suggestions. Additionally, Copilot's ability to create and manage checklists in Planner is limited, which can be frustrating for users who rely on this feature for task management. Copilot's performance can also be inconsistent when it comes to scheduling meetings, as it may not always accurately find suitable times or integrate well with calendar data. These limitations highlight the importance of understanding Copilot's capabilities and setting realistic expectations for its use. Like humans, Copilot is very good at some things and less good at others. Some staff gave it a name which helped them to forgive its weaknesses as it continues to learn.
Whether you are successful or not with your prompts, it’s important to remember that Copilot is evolving every day and things that may not work so well one week, might be better next time – don't give up because it didn’t work as you expected the first time.
Next Steps
Copilot is still clearly in development and IT Services staff are providing feedback to Microsoft about potential improvements or bugs they come across. But nonetheless, the IT Services Copilot trial has demonstrated the potential of AI to enhance efficiency and collaboration within the department.
Our focus continues to be upskilling colleagues with a view to Copilot becoming art of the department’s ‘business as usual’ provision in the very near future. Stuart Lee commented:
Slowly but surely, staff members are becoming increasingly familiar with Copilot AI to speed through certain tasks giving them more time for other work or just to take time out to remove the stress of the day.”
Explore Copilot for yourself
If you are lucky enough to have a Copilot license but don't know where to begin, explore the below links as a starting point to see what's possible and help create prompts so that you can learn and get the most out of this useful resource. You’ll need to invest a little time in learning how to create awesome prompts, but it will be worth it in the long run.
Licenses can be obtained from the AI and Machine Learning Competency Centre, which also provides wider support and guidance on generative AI tools.
Why not give it a go, just click on the Copilot icon and make your request!
Useful links for getting started with Copilot
Microsoft guidance
- Microsoft Copilot video tutorials - Microsoft Support
- Copilot Prompt Gallery
- Manage Microsoft 365 Copilot in Teams meetings and events
University resources
- Find AI courses (Search for 'AI' in the resource finder for courses and resources about using AI at the University.)
- Get help and expertise from the University's AI and ML Competency Centre
- New hub launched to explore use of artificial intelligence in education (Article 2024)
Getting a Copilot license
What others are saying
- Digital strides (Article 2024)
- Microsoft case study about IT Services (Article 2024)
- AI in Oxford: experiments, tools and ways of working (Article 2023)
