The Expert View: Transforming AP and AR with AI, without growing the finance team

Finance workloads keep rising. Finance teams, for the most part, do not.

 

That tension ran through a Business Reporter dinner briefing at the House of Lords, hosted by Quadient, where senior finance leaders discussed whether AI can absorb the growing demands of accounts payable (AP) and accounts receivable (AR) without adding headcount. The technology, attendees concluded, is largely ready. The harder question is whether people are ready to trust it.

 

Opening the discussion, Dominic Febers, Finance Automation Expert at Quadient, set out the case for doing more with less. “Finance workloads are increasing, but finance teams are not expanding,” he said. AI can take on the manual work in AP and AR, he argued, and that gives people time to focus on the work that adds real value.

 

What counts as AI?

 

The group began by discussing use cases, such as those in construction – a sector that remains stubbornly traditional. Contracts can run to several thousand pages, and AI can review them in a fraction of the time a person needs. However, another participant had fed one contract to two different AI tools and got two different answers. She talked about losing trust in a tool “when I know it’s wrong and it insists it’s correct”.

 

The remedy, some said, is to treat AI like an employee: give it good instructions and check its work. The tools need security around them and their output needs verifying. One warning note: automation of any kind is a challenge for inexperienced staff, because they do not know the underlying process well enough to spot when something has gone wrong.

 

Replace or refocus people?

 

Some jobs will be replaced by AI, attendees accepted, but new roles will appear too, including people whose job is to manage AI output. Roles are already changing shape, one participant said, with order takers becoming “proactive sales makers” as AI takes on the routine end of the job.

 

The deeper question was what to do with the time AI frees up. For most, this was about creating value: if AI automates order processing, people can be redeployed to improve customer service. However, trust surfaced again, with one attendee saying they could replace people with AI in two areas of their business today, “but do I trust the AI to be more accurate than people?”

 

The pace of change worried the room more than the direction. AI will change the workforce and people will do different things, attendees agreed, but it is happening at a speed organisations – and the education system – are struggling to absorb.

 

Chasing payments at machine speed

 

On the AR side, the discussion turned to a current trend: AI bots calling customers to chase payments. Whether that is welcome depends on the customer, attendees said. It will not work for customers who need empathy, such as small businesses struggling to juggle finances, or for big accounts that need careful management. But customers are getting used to talking to bots in their consumer lives, one pointed out, and will get used to it at work too.

 

The subtler application drew more enthusiasm. AI is good at finding the customers who are low effort and high value when it comes to paying bills. And if the data shows that a phone call gets an invoice paid in 45 days rather than 55, the system can schedule a person to make that call. Speed and cost effectiveness were the priorities the group kept returning to, along with reducing errors, and AI can check for those too.

 

Buy, build or partner

 

Adoption pressure is coming from several directions. Board-level demands to transform finance with AI often turn out to be cost-cutting pressure, one attendee suggested, or a push for AI as a route to efficiency and revenue growth. Meanwhile, employees are using AI in their personal lives and want to use it to transform their experience at work.

 

Attendees also said that AI brought a new twist on the question of whether to buy or build your technology. One was costing up whether to buy an OCR tool or use Claude to build one in-house. Attendees leaned towards buying: an off-the-shelf product is known to work and has governance built-in, while your own tool might work inconsistently or introduce governance risks.

 

There is real risk, another attendee added, in spending money to build a tool that fails. The ideal, some felt, is a bespoke tool developed with a supplier that understands the business. Two things would speed adoption: trial periods, so buyers can assess the benefit before committing; and proper change management, which matters as much as the service itself.

 

Closing the discussion, Febers returned to the theme of trust. “You can sense that AI is the main topic in finance right now, and everyone is at a different point of the journey,” he said. The challenge organisations share is trust, he argued, and the answer is unglamorous: governance, verification and suppliers who understand the business. Get that right, and finance teams really can do more without growing.

---

To learn more, please visit: www.quadient.com