Signing up to an AI platform has never been easier. It takes employees a matter of minutes, often with little more than a free trial or subscription charge to a company card. And while businesses are racing to embrace AI, in many cases they haven’t made secure, approved tools available at the same pace.
The result is a growing blind spot for procurement teams. Our research found that 27% of employees surveyed had purchased AI tools without approval, and it’s not difficult to understand why. Employees may be frustrated by lengthy approval processes or want to demonstrate what a tool can do before making a formal request.
But what starts as a small, seemingly harmless purchase can quickly become a bigger business challenge. Different departments may subscribe to overlapping tools, recurring payments can go unnoticed and usage-based charges can grow well beyond the initial cost. Without clear visibility over these purchases, spending becomes increasingly difficult to track, let alone control. By the time costs appear in month-end reporting, procurement and finance teams are left working backwards to establish what was bought, who is using it and whether it is delivering value.
For CPOs, the challenge goes beyond managing the cost of AI adoption. They're also being tasked with using AI to improve procurement itself. But without a reliable picture of where money is being spent today, how can they make informed decisions about where AI could deliver the greatest value tomorrow?
Gain greater visibility into spending
Finding out where AI spend already sits is the first step to creating a more reliable overview. This includes looking beyond centrally managed contracts and bringing together purchases made through company cards, expenses, procurement systems and supplier invoices. That means asking: Which tools are people using? Who bought them, and for what purpose? Are different teams paying for similar services?
It’s important to understand that the answers may also expose risks beyond cost. Someone experimenting with an unapproved tool could enter company or customer information without the business having checked how that data will be stored or used. Procurement departments need to know which tools are in use to be able to work with IT and other relevant teams to assess them properly.
A more accurate spending picture can also help CPOs identify where AI might improve procurement itself. Consider a team using AI to look for overlapping suppliers or subscriptions. If its analysis draws only on centrally managed contracts, it could miss purchases made elsewhere in the business. And if transactions lack information about their purpose or owner, apparently similar purchases may be difficult to assess.
Spending data is only part of the picture; contracts and supplier information are also key. The point is that AI can make patterns easier to spot, but it cannot make a sound recommendation from information the business has never captured.
Aligning purchases with process
With a clearer view of the different types of AI spend, CPOs can put the right controls around them. The aim should be to move governance closer to the point of purchase, rather than discovering a problem weeks later. However, that doesn’t mean sending every tool through the same approval process.
For example, large, strategic purchases generally already follow established procurement routes, from supplier selection and negotiation through to approval and payment. Alongside them is a long tail of smaller transactions, including software subscriptions and tools that employees want to test before making a bigger commitment.
A low-cost trial and a tool that will process sensitive information or be rolled out across the business should not necessarily face the same level of scrutiny. Spending limits and approval routes can reflect the value, purpose and risk of a purchase. They give employees a way to explore new tools while keeping procurement aware of what is being bought.
Centralising every decision risks making the process so slow that useful experimentation stalls or moves out of sight. A better approach is to decide which purchases can be delegated, under what conditions, and when a closer review is needed.
Make approved purchasing straightforward
Having rules in place will only work if people understand them and can get a timely answer. Our research found that 60% of employees surveyed said they frequently bend rules or find loopholes to access company money, while 29% said finance processes had delayed or slowed their work. If the approved route cannot keep up with a genuine business need, people might look elsewhere.
Instead, procurement can provide teams with clear options for trying approved tools, with a route for requesting something new. Capturing the purpose, owner and relevant spending information when a purchase happens means the business does not have to reconstruct it afterwards. It also gives procurement a clearer view of demand, even when decisions are made across different teams.
This is a question of capability, in addition to access. Employees need to understand what information is safe to put into a tool. Procurement teams using AI in their own work need to know what data it has drawn on, where gaps might exist and when a result requires human judgement. A recommendation that looks convincing still needs to be checked against supplier agreements and genuine business needs.
Building a more strategic approach to AI
Getting AI adoption right doesn't mean investing in every new tool or application that comes along. For procurement teams, the priority should be building a clearer understanding of existing spending, making it easier for employees to experiment with approved tools and agreeing how to assess usage, costs and the value each tool delivers.
This also means recognising that AI adoption is an ongoing process, rather than a one-off investment decision. As new tools emerge and business needs evolve, CPOs must be prepared to reassess which solutions are delivering value and where a different approach may be needed.
Ultimately, procurement has an opportunity to play a much more strategic role in how businesses adopt AI. By combining greater spending visibility with clearer purchasing processes and a more consistent approach to measuring outcomes, CPOs can move beyond simply controlling costs. Instead, they can help the wider business make more informed investment decisions, ensuring AI adoption is driven by genuine business needs and delivers meaningful, long-term value.