We are thinking of doing a pilot of CoPilot, any recommendations in terms of helping the user community unlock the most valuable features of CoPilot?
Sort by:
Gartner ran a really good presentation last October with some solid metrics & insight around CoPilot rollouts: "GTP Client Webinar: Copilot for Microsoft 365: Assessing the Impact and Value So Far"
I highly recommend talking to your Gartner account team to see if you can get access to the recording. There are some important lessons learned by others in there to be aware of- Specifically about the efficacy of rolling CoPilot out to users en masse.
I am running our "pilot" program for Copilot. It has been going well. I recommend enabling user communities rather than single users. And those communities should be asked to articulate anticipated outcomes.
The users of Copilot should know a bit about prompt engineering (How to generate a good prompt) and also recognize that not all Copilots are the same. From what we have experienced, Teams Copilot appears to be a different model and/or has access to different context data (e-mails, Team meetings, etc.) than Word Copilot. Copilot's ability so summarize and condense data and e-mail threads is fantastic. Its ability (in Teams Copilot) to generate potential action narratives based on encountered situations is also something they should look at.
Can you please share why enabling communities is superior to enabling single users.
Hello James, thank you for the question. For all I know enabling communities rather than single users is a terrible idea. I assume that it depends on who is in the community. But our thinking on the idea is that whoever is engaging these LLMs are less likely to do so dangerously and more likely to have support and input from collaborators. It is not a "Safety in numbers" argument but more "iron sharpening iron". Copilot is fairly "safe" in our use of it so far, but there IS some effort required to get the most benefit out of it. We consider that having people working collaboratively in communities will both decrease that ramp time and increase the likelihood that value will be realized.
Hi, we ran a pilot last year and I'd start with your objectives. We went with
* To build digital adoption and engagement with Copilot for M365 for trial participants
* Measure impact on productivity, creativity and wellbeing for trial participants
* Inform the business case for further roll out of Copilot for M365 licence
We then baselined our adoption readiness - current usage, change impact, upskilling approaches and expected outcomes as well as our governance and technical readiness.
During the trial we tracked the benefits people achieved and positive/negative sentiments, time saving potential, impact (especially in content generation), community learning, the use of copilot chat specifically, and also our adoption activities.
We found adoption activies were one key area and these were weekly online webinars and round tables, champion groups, regular email communications and Viva/Yammer communities.
One key area before you start is to map out where your data repositories that Copilot will interrogate are located. If they are in OneDrive or SharePoint Online, you're good to go. Copilot will not use data on legacy file & print servers/network shares or other offline stores. Also, if your users base is used to using desktop MS apps, then you will need to upgrade them to later versions of M365 Desktop Apps.
Office 365 Co-pilot or Github Co-pilot? If the latter, I would recommend Cursor above Co-Pilot
If GitHub Copilot, I would suggest the official learning path and exam: https://examregistration.github.com/certification/COPILOT
Which part of copilot. Microsoft has used copilot as a branding item to cover all AI capabilities in all its products. Before launching any GenAI, be sure to have policy in place and ensure your data rights and permission are robust.