The Easy Part: Creating the Asset
Across the higher education landscape, universities are adopting AI tools for operational efficiency to help current staff do more within their current bandwidth. AI can speed up a lot of work: code development, data analysis, email generation and the list goes on. One functionality we’ve been exploring more deliberately is asset creation. In our work, we rely on several external tools for data collection, storage, analysis and visualization: Qualtrics for surveys, a data warehouse for storage, Tableau for visualization. With AI, we can now create dashboards like these with a clear prompt, without looping in multiple teams or specialty roles. AI helps us build the dashboard, and we maintain it simply by uploading new data on a regular cadence.
We’ve applied the same approach to market and competitive intelligence: dashboards that update themselves with the latest news, market trends and activity relevant to our work. Rather than someone manually checking a handful of news sites and publications every week, we prompt AI to search for relevant trends, such as new competitor activity, shifts in demand, major hires or departures, funding news and announcements that signal where the space is heading. We give it a clear set of requirements up front, so we know it’s pulling from current, reliable sources.
From this input, AI reads through everything it finds, pulls out the relevant pieces and organizes them into a clean layout with clear headers by topic and urgency, so it’s obvious at a glance what’s a general trend versus something that needs immediate attention. It also generates a short set of action items for each finding, including why it matters, what it might mean for our work and a recommended next step, like reaching out to a specific contact or revisiting a stalled conversation. The dashboard refreshes on a set cadence through weekly prompting, so the information, recommendations and prospects it surfaces are always current.
We determined that this same setup scales down as easily as it scales up. If we want a wide view of an industry, we point AI at the market. If we want a focused view of one organization, we point it at that single company, so it pulls press releases, leadership changes and recent announcements specific to it. It’s the same core process, just applied at a different scale, which turns a task that used to take real time and manual effort into something that updates itself on its own.
So, creating the asset ended up being the easy part. Sharing it reliably, continuously and without creating a maintenance burden turned out to be the hard part. That’s the last-mile problem this piece is really about.
Where It Broke Down: Sharing the Asset
AI gave us complex code files that could be shared and opened locally, but updating and redistributing those files on a regular basis, as we improved the backend, became difficult to manage and frustrating for stakeholders, as nobody could be sure they were looking at the most current version.
It’s easy to build an AI dashboard in any department. Anyone can take some data (anonymized, no personal information), upload it to an AI and ask for a Tableau-like dashboard of the analytics. You don’t even need to know exactly what you want. The AI gives you a first draft, and you iterate from there. A skilled data analyst will do this faster and better, but a novice can spin up something useful when time is tight. The problem is what happens next.
That dashboard lives in a personal AI account with no sustainable way to share it externally. We felt this acutely on our cross-functional team, which spans more than 50 members across several roles and reporting lines. At that scale, informal sharing methods buckle as assets get lost in the shuffle, and it’s never clear who has the latest version. It got more complicated because we regularly share assets with teams elsewhere in the university, outside our direct group, which meant we couldn’t rely on any shortcuts specific to our own team’s habits or tools.
The obvious fixes weren’t good enough:
Emailing the code file: Fine for a one-time handoff, but every backend update meant a new email and version control fell apart almost immediately.
A shared online folder: Better than email, since at least the file lived in one place, but it still required the end user to download and open code locally, with several points where that could go wrong. More cumbersome than it should have been for what was meant to be a simple handoff.
The Fix: Finding the Right Tool Already on Campus
Both solutions above were inefficient and error-prone, so we set out to research what tools might already exist within the university that could solve this problem. We looked at a range of options, including a GitHub repository, HTML hosting and a dedicated departmental server, which would have required additional investment. These fixes were all more technically advanced than the problem called for. We didn’t need that level of infrastructure; we just needed one link that reliably served the current version of the asset, with the ability to update the backend without asking end users to do anything differently.
That research led us to CampusPress, a WordPress derivative with pre-approved university branding and templates, already sanctioned for use across campus. It turned out to be exactly the right fit for this use case, and it checked every box we needed:
- Easy share-out: The asset would reside within this one link always
- Easy backend updates: We can swap out the underlying code without the end user ever knowing, and they will land on the current version.
- No new approvals: It’s already a university-approved tool, with branding and security built in.
What We Learned and What Others Can Take From It
AI turned out to be more operationally impactful than we initially expected, but realizing that potential took more than good prompting. It required an underlying foundation for broad share-out, so the benefit could reach everyone who needed it, not just the person who built the dashboard.
If your team is running into a similar wall, a few questions are worth asking before you invest in a fix:
Is the bottleneck creation or distribution?
These are different problems with different solutions. If your team can build the asset in an afternoon but can’t keep it current for stakeholders, there is a distribution problem. Therefore, there is no need to over-invest in better prompting when the fix is downstream.
Does your institution already have an approved, low-lift hosting tool?
Before exploring custom infrastructure, ask IT what’s already sanctioned. We assumed we’d need something built for us, but it turned out a tool already existed that just needed to be pointed at the right use case.
Can the backend update without end-user action?
This is the real test of a sustainable solution. If updating the asset requires anyone downstream to re-download, re-open or re-request a file, the system will eventually fail at scale.
Will this still work outside your team?
Solutions that rely on your team’s specific habits or tool stack tend to break down the moment you share with a different group. Build for the recipient with the least technical context.
AI can make asset creation mostly frictionless, but that only translates into organizational value if the last mile—getting the asset reliably in front of the people who need it—gets the same attention as the asset itself.