Part 1 – Meet the Analytics Solution Consultant (and a Quick Primer on Copilot Agents)
If your work at a Microsoft institution, then this post is for you. And if you do analytics work in higher ed, you probably have more questions than tools to answer them, and more tools than people to run them. Over the next three posts I want to walk through three Copilot agents that can help with exactly that kind of problem. The nice part is they’re built on things most of us already have sitting in your Microsoft space. Let’s start with what a Copilot agent actually is.
So what’s a Copilot agent, really?
Think of it as a focused version of Microsoft Copilot that you point at your own content and give a specific job. Instead of generic chat, it answers from the documents you feed it and talks the way you tell it to. The lightweight ones you can build right inside Microsoft 365 Copilot Chat (or Teams) with the Agent Builder. There is no code, just plain instructions and a knowledge source, which is usually a SharePoint site where your docs live.
The part everyone asks about: what access do I need?
As of mid-2026 (Microsoft moves this stuff around constantly, so confirm against your own tenant before you commit):
• To build an agent grounded in your own data (SharePoint, Graph), you will need a Microsoft 365 Copilot license OR have a pay-as-you-go Copilot Studio metering turned on. Check with your institution's IT department for details on pricing. If you already have Copilot, you’re good: the standard license now includes Copilot Studio access. Knowledge sources (Microsoft Learn): states plainly that SharePoint and OneDrive knowledge sources require the signed-in user to have an active Microsoft 365 Copilot license, and that requests fail without one.
• Agents respect existing permissions. The agent only ever surfaces what a given user can already open in SharePoint.
• To use an agent, people need at least Copilot Chat (bundled with most M365 plans). Usage against your tenant data is covered either by their Copilot license or by Copilot Credits.
• Make sure that your M365 admin role gives you Copilot access and agent creation, and that you have Power Platform admin rights as well (Copilot Studio environments, connectors, DLP).
That’s the permissions related stuff. Now the fun part! Your first agent. Something you can try out if you don’t have a clue where to start.
The Analytics Solution Consultant
Every analytics lead eventually hits the same wall: which tool for which job, and who do I need to run it? Dashboards versus a warehouse versus self-serve versus predictive modeling, and then team structure. The answers usually live in a few people’s heads and a scattering of old emails.
The Analytics Solution Consultant agent is a sounding board for those decisions. You feed it your tool documentation, your license and cost info, and your team role descriptions, and it gives a fast first opinion that’s grounded in your institution’s actual constraints and not a generic internet answer. It won’t replace your judgment and you will most likely need to iterate several times to get it close to what you need. But it will give you something to work from in minutes instead of weeks, and it keeps everyone working from the same playbook.
Here are a few sample questions to feed it:
• Which tool should we use for executive dashboards versus ad-hoc exploratory analysis?
• For predictive modeling on large datasets, what platform do you recommend given our budget and the team’s skill level?
• How should I structure my analytics team, and what roles do I need to cover reporting, data engineering, and data science?
• What are the trade-offs between two specific platforms for self-serve analytics?
To set it up: create a SharePoint site with that documentation, point the agent at it, and give it simple guardrails. Remember to be concise, refer to shared docs, and defer to the data and analytics governance team when unsure. That’s it.
One thing to be clear-eyed about: this agent is only as smart as the material behind it. It doesn’t invent your tool strategy but it makes the strategy you’ve written down easy to consult. If your SharePoint site is thin, the agent is thin.
I put this one first on purpose, because it’s the most strategic of the three. The next two get hands-on, and the first of those is the one I think hits closest to home for HEUG folks. In the next post we’ll finally document that PeopleSoft environment, without buying a cataloging tool.