Getting the most out of Microsoft Copilot

Overall summary

This session wasn't about new technology. It was about getting more out of a tool most campuses already have.

Microsoft Copilot ships inside the Office 365 suite that the majority of institutions already license. But for most staff, it sits unused, or gets treated as another chat window. The session walked through three practical ways to put it to work: automating role-specific research, interpreting data without a data analyst, and generating custom images. None of them require technical skill. All of them can be tried this afternoon.

Staff joined from across eight to ten provinces and states — VPs of student affairs, instructors, directors of student success, facilities management, even parking and transportation. The breadth made the point on its own: this is not a conversation for technical people. It is a conversation for anyone with a job to do.

The main idea

If the session had one takeaway, it was this:

You no longer need to be good at prompting. You need to be good at describing the work.

The skill of writing clever prompts is fading in importance, because AI will now write the prompt for you if you can explain what you want — the way you would brief a colleague or someone you're training. Each of the three use cases below comes back to that shift. The work moves upstream: from typing instructions to describing outcomes.

Use case one: Research that runs on a schedule

The old workflow for staying current — job postings, grant releases, legislative changes, competitor programs — is one everyone recognized: Google searches, a dozen open tabs, newsletter subscriptions, and results that depend entirely on the time you put in.

Copilot has a feature called Tasks that changes the model. You write one prompt — "monitor X and summarize it for me every Monday at 9am" — and Copilot runs the search on schedule and delivers something closer to a personalized newsletter built around your role. It is a first step from chatting with AI to giving AI a job.

One attendee in recruitment saw an immediate fit: his team's annual competitor analysis, normally weeks of manually combing through other institutions' websites for tuition and admission changes, could become a standing task.

Use case two: Interpreting data without a data analyst

The most resonant story of the session involved a pilot program at a Michigan campus. The dean called the student data it generated "a treasure chest." But interpreting thousands of rows of survey responses meant borrowing scarce time from IT, and in the end, nothing came of it. The pilot ran; the insight stayed locked in a CSV.

That problem has quietly disappeared. Copilot now lives inside Excel, and the prompt that produced a working executive dashboard during the session was, in full: "Can you make me a dashboard on the raw data table?" A minute or two later, the dashboard existed in a new tab.

Data scientists joke that 80% of the job is cleaning data before any insight appears. That 80% is what just got automated. But the session was equally clear about the boundaries:

• Whatever Copilot produces is a first draft, not a final say. Its job is to get you asking the right questions, not to replace analysis.
• Deeper work still belongs with a real analyst — the difference is that the analyst now spends their time on work that actually needs them.
• Keep it simple. One table at a time, raw data before charts. Too much context degrades the output.

The session also named the discomfort directly: a dashboard that once required hiring someone now takes one sentence. That raises a fair question about what happens to that work. One attendee put the answer well — AI makes the process faster, but it doesn't eliminate the need for the humans who understand the context and know what the data means for their campus. The change is to tasks, not to the need for people.

On data security, the guidance was plain: logged in through your institution's enterprise account, Microsoft commits to not training models on your data. Outside that license, the protections don't apply. And for anything involving student personal or health information, check with your IT department first.

Use case three: Custom images without the stock-photo hunt

Copilot generates custom images, including directly inside PowerPoint, where it can read the slide and place the image for you. Marketing assets, event branding, presentation visuals — work that used to mean scouring Google Images or paying for stock now takes a prompt.

The more useful lesson was a technique: don't write the image prompt yourself. Ask AI to write it first. Describe what you're trying to create, ask it to ask clarifying questions, then use the detailed prompt it produces. One attendee called the approach a game changer — the AI came back asking whether he meant mint green or grass green, a level of specificity nobody includes on their own.

The same pattern extends beyond images: context first, then the task. Some of the best results come from talking through the project with AI conversationally — even by voice — before asking it to produce anything.

The adoption question

One attendee raised the hardest question of the session: the interfaces keep changing. A feature that worked two weeks ago disappears or moves, and for staff already hesitant about AI, every change is an excuse to disengage.

The honest answer is that this will stay messy for a while. The industry is still working out its business models, and features will keep shifting. The practical response is to teach durable, task-level skills rather than chasing whatever launched this month. A dashboard from a CSV isn't cutting-edge — it's just useful. And useful is what gets a skeptical colleague to lean in.

That is also the reason these sessions exist. AI learning should be accessible to everyone on campus, not just the people with the title or the budget. Tailwind hosts a free public webinar roughly every month; the next topic will be shared by email, and suggestions are welcome.

Access to Recording

To watch this webinar, please email devin@teamtailwind.com for access.

Frequently asked questions

What is Microsoft Copilot Tasks?

Tasks is a Copilot feature that runs a prompt on a recurring schedule — for example, monitoring news or program changes and delivering a summary every Monday morning. It began rolling out to enterprise suites in 2026 and may not be available in every tenant yet. It is a simple first step into automation: instead of chatting with AI, you assign it a recurring job.

Is it safe to put college data into Copilot?

Under an institutional enterprise license, Microsoft commits to not using your data to train its models, and Copilot use should fall within your institution's policies. Those protections do not apply to personal Copilot accounts, so always log in with your institutional email. For student personal or health information, confirm with your IT department before uploading anything.

Do I need to be good at prompting to use Copilot?

No — and that skill matters less every month. The more durable skill is describing what you want clearly, the way you would brief a person you're training. A practical technique: ask AI to write the prompt for you, and have it ask clarifying questions first. The dashboard shown in this session came from a single plain-English sentence.