)
How I’m repositioning Spendesk to meet our clients’ growing desire for AI tools
For all the excitement about generative AI since ChatGPT launched three years ago, finance teams have understandably been among the slowest adopters. When dealing with money, no one wants to risk an AI hallucination.
Thankfully, the technology is finally maturing, and that conservatism from finance teams is steadily thawing. Talking to Spendesk’s clients, it’s clear that there’s been a vibe shift recently. A typical SME client might have a five- or maybe ten-person finance team, whose bandwidth is 80% taken up with tasks that are backward-looking. From these kinds of teams, who want to free up their time for tasks that are forward-looking and strategic, I’m increasingly hearing “I need AI” or “AI is the solution to my problem.”
I’m already repositioning Spendesk to meet this growing client demand for AI tools. Let me walk you through where I currently see the most opportunity.
For a start, let’s think about fraud protection. At the moment, finance teams have to create manual workflows for flagging potentially fraudulent transactions for review. You might say, only alert me if an employee below this level of seniority purchases something for more than €50. It’s fine, but it’s not exactly smart. With AI, we can build agents that mine huge pools of data to predict the chance of fraud in any single transaction. We can program these agents to only flag something if it’s suspicious. I don’t want to be told about purchases where there’s a 99.99% chance that everything is normal.
Secondly, let’s think about the “procure to pay” process, where AI has huge potential to parallelise tasks and bring efficiency. Currently, a purchase order may need to go through half a dozen reviews before it can be approved: a legal review, a security review, a compliance review, a finance review, etc. While each of these reviews may only take 2-3 hours of someone’s time, one team won’t make a start until the previous step in the process has been signed off. Running these approvals one after the other means that the time taken from start to end can bloat to maybe 20 days or more – plus each step has to be chased by members of the finance team, playing bad cop. It’s hardly rewarding work.
AI holds the promise to convert this serial process into a parallel one, where all the various decisions can be taken concurrently. This should be able to collapse the process from weeks to just a few hours.
The Spendesk suite
Spendesk gives you visibility on all the money flowing out of your business. It’s a powerful suite of products, but there’s so much more to come.
The next iteration of Spendesk – the one that we’re building at the moment – will give much more context and allow clients to make smarter decisions. It will be able to look across huge pools of data and quickly answer questions like “is the expense you’re about to approve within budget?” or “Are you paying the right price for this service?” We’ll be able to bring in competitive intelligence, looking at what other businesses of your size pay for products like this. These next-generation tools will allow you to spend smarter and ultimately optimise your P&L.
As a tech CEO, I’m of course excited by the latest technology. But, when it comes to AI, I’m mainly excited by the real-world problems these tools can solve for our clients.
When Spendesk launched in 2016, our initial product had an undoubted “wow” factor. As we move into 2026, I’m confident there will be many more wow moments to come from Spendesk.
)
)
)
)
)