Before You Buy Another AI Tool, Check What You’re Already Paying For
Originally published on November 16, 2026
For years I have watched institutions underuse the systems they already own. Nearly every client I have supported from a technology standpoint has an ERP capable of far more than it is being asked to do, and while that has started to improve, the arrival of AI has the same pattern repeating at a faster pace.
The sequence is familiar by now. A new AI tool is advertised, someone on campus sees a demo that looks great, and a department finds the budget. Six months later, the institution is paying for a capability it already had, or was about to get, from a system it already licenses.
Two ways the money gets spent twice
When I look at a proposed AI purchase, I am less interested in whether the tool works, because most of them do, than in what it does that the institution cannot already do. Two answers come up often.
The tool adds only incremental value over the general-purpose assistant. Most campuses now license ChatGPT, Claude, Copilot, or some combination of the three, and the core capabilities of those assistants cover a large share of what many specialized products demonstrate. The specialized product may deliver it through a more polished interface, but that alone rarely justifies a second contract.
The system that houses the data already has AI built in, or soon will. Many ERP vendors are rolling out AI capabilities now. When the AI sits in the same system as the data, it is typically cheaper and more secure, and it is likely to be more accurate and reliable than a separate tool working at a distance from the source.
The second point matters most for finance leaders, because the value of AI in a business office depends on the data it can reach. The vendor that already holds your general ledger, HR records, or student accounts starts with an advantage that a newcomer can only match through an integration you end up paying for.
The cost that is not on the invoice
There is a second cost that rarely makes it into the purchasing decision, and as a former auditor it is the one I notice first. In financial accounting, reporting, and internal controls, what matters is the trail: the documentation behind every adjustment and the support for every balance. Any time a new technology is added to that flow, it adds the risk of transformations or adjustments that are not adequately supported.
That risk can be managed, but not for free. Every additional tool sitting between the source transaction and the reported number is something your auditors have to understand and test, and something your staff have to document and explain. A tool that saves time during the close can end up costing time, effort, and fees during the audit, and that trade is seldom part of the business case.
The procurement clock
Then there is the matter of time. I have seen institutions find a new vendor whose solution is capable and would add real value, only to have the process of bringing that vendor on take longer than it would have taken to get near-term value from a partial solution built with the tools they already had.
A partial solution can sound like a compromise. With AI capabilities growing as quickly as they are, it is often the best end result anyway, because the tools it was built on keep improving in the meantime.
Why it keeps happening
In my experience, this comes from four ordinary conditions.
- Nobody knows what shipped. New capabilities arrive in existing platforms on the vendor’s schedule, and most of the people who would benefit from them never hear that they exist.
- IT review takes time. A new feature inside a licensed system still has to be reviewed and approved before it is turned on, and when that does not happen quickly, a department with a problem to solve will not always wait.
- Vendors under-explain. Even when a capability is available, vendors do not always do a good job of explaining when and how it should be used.
- The new tool is compelling. A focused demo from a company that does one thing well will often look better than a feature tucked inside a system people associate with month-end close.
The available evidence suggests the problem is broad. Campus Technology recently reported on one campus that found 11 different enterprise project management tools in use and another that identified dozens of CRM instances. Outside higher education, Zylo’s 2026 SaaS Management Index found that the average organization uses just 54% of the SaaS licenses it pays for. In a 2025 EDUCAUSE QuickPoll, only 36% of respondents whose institutions had an AI procurement process said AI products are reviewed again after the initial purchase.
That last number deserves attention. Because AI capabilities inside existing platforms are changing from quarter to quarter, a purchase that made sense when it was approved may be redundant a year later, and most institutions are not going back to check.
Where I would look first
From a finance and audit perspective, I would start with the data layer rather than the application layer. If your institution runs a platform such as Snowflake or Databricks, it already has strong AI capabilities that are only going to grow, and because that is where your data is consolidated and governed, it is the logical first place to ask what is possible before evaluating anything new.
From there, the same question applies to the ERP, the student system, and the productivity suite: what is licensed, what is turned on, and what is on the roadmap.
When a standalone tool is the right answer
None of this is an argument against buying AI tools. A standalone product can make sense when the use case it was built and trained for lines up directly with what you need it to do.
Grants management is a good example. A tool built specifically around the portions of grants management that are a real challenge for your organization may do that work better than a general assistant or an ERP feature ever will. The test is fit: a tool built for your problem can earn its place, while a tool that needs you to find a problem for it usually does not.
Three questions, three owners
Before the next AI purchase, an institution should be able to answer three questions:
- What is available to us today in the systems we already pay for?
- What is coming soon from those same vendors?
- What are the gaps that neither will close?
The difficulty is that no single office holds all three answers. IT knows what is licensed and what has been approved, Procurement knows the contracts and the renewal dates, and program and finance leadership know which problems are worth solving. Keeping that picture current has to be a joint effort among all three, and it has to be maintained, because the answers will keep changing.
If your institution cannot answer those three questions today, that is the finding, and closing that gap is a better use of the next ninety days than another demo.
Before investing in another AI solution, make sure you’re getting the most from the systems you already have. James Moore Digital helps institutions look beyond the latest AI demo to identify where technology can deliver meaningful value, with support ranging from evaluating existing capabilities and identifying opportunities for automation to AI implementation and governance.
All content provided in this article is for informational purposes only. Matters discussed in this article are subject to change. For up-to-date information on this subject please contact a James Moore professional. James Moore will not be held responsible for any claim, loss, damage or inconvenience caused as a result of any information within these pages or any information accessed through this site.
Other Posts You Might Like