Where Automation Works and Human Judgment Still Matters

0:01I think the speed is changing the tone in which where, hey, if approvals used to take 50 hours, now it's taking seven hours. Finance really stops being that bottleneck or the workaround people see. It's a relationship change and not necessarily a process change.

0:20Megan Weis: Welcome back to CFO Weekly, where we're talking with financial leaders about how to build efficiency in their teams, create time for strategy, and ultimately get results. This podcast is brought to you by Persona, the trusted leader in finance and accounting outsourcing for over 30 years. See how Persona's customized solutions can help you streamline your operations with teams that start as small as one. Visit the website at persona.com to learn more. I'm your host, Megan Weiss. Let's jump right in.

0:54Megan Weis: Welcome back to CFO Weekly. Today, I'm joined by Amy Wang, Chief Financial Officer at Procurify, an AI-enhanced spend management platform helping mid-market organizations gain greater visibility and control over their financial operations. Amy's experience spans high growth, SaaS, fintech, and capital markets, including leadership roles at Morgan Stanley, Solium, Userful, and now Procurify, where she oversees not only finance, but also people operations, legal, IT, and procurement. Amy is deeply focused on helping organizations build finance functions that are disciplined without becoming bureaucratic, and modern without chasing hype.

1:35Megan Weis: In this episode, we'll explore what AI and finance actually looks like inside mid-market companies, where automation genuinely adds value, where human judgment still matters most, and how finance leaders can make smart technology decisions in a landscape that changes almost every six months. Welcome to this episode, Amy, and thank you so much for taking the time to be here with me. Thanks, Megan. I'm really glad to be here. So to start, can you walk us through your career journey and how it shaped the way you think about technology, operations, and modern finance leadership? So in terms of my own career journey, it has definitely been a bit of a tour from a lot of different angles in terms of finance and accounting.

2:19I had started out my career at PwC. What I'd like to say is I did my time in audit and insurance. But joking aside, I did really enjoy my time there. Not only did I learn a lot, but what I got to see was a lot of range. So auditing and working with a lot of different businesses from the likes of oil and gas, mom and pop shops, a lot of tech businesses, as well as a lot of financial services. So you kind of started to see like the same pain points over and over, that same manual work, reconciliations, all the stuff that our teams have been consistently absorbing over time. So not long after PwC, I decided to move into an industry role to a company called Solium.

3:07Their fintech had really a good product in terms of stock-based compensation management. And that's kind of where I put on my operator gloves, really learning the business from the ground up. And it's really where I fell in love with product-first companies. And so for me, it's when the product leads and their buyers are that finance team, you really get to have a lot different impact and input into the business itself. So at Solium, I grew into leadership roles, worked through global compliance, money movement, treasury reporting, the like. But I would say the highlight of that role was really helping the company get to its billion dollar exit over to Morgan Stanley.

3:54and that's where I had stayed on at Morgan Stanley after the acquisition, which was its own education and experience. I think that's where I really saw behind the scenes on what enterprise really looks like. A lot of red tape and bureaucracy, but from a technology standpoint, really outdated technology. You would be surprised in terms of what they're working with, but really like very smart people that had to work around that every day. And from there, I took my first CFO role, Userful, where I was able to help own the full function in a growing tech company and helping really advise and build on what the next stage of investment and technology look like.

4:40And that's where it's ultimately brought me to Procurify today, where, again, my passion is helping improve the pain points for our finance team. And so seeing what Procurify had to offer in terms of being a product first company and what was ultimately doing for the office CFO, I wanted to be on that side in terms of building for that fix. And that's where Procurify really sits at that intersection, right, in terms of spend management. We're very much built for the mid-markets. And really, these teams feel the pain points most and have the least capital and ability or resources to absorb those pain points.

5:23So I think to wrap it up, I've really seen finance from all angles, from that auditor chair to scale up and really inside that enterprise and now from a CFO lens. And that's helped me really think about technology differently, about what tool stack I recommend or bring into an organization, and then thinking about how I really measure my decisions and investments long term.

5:48Megan Weis: Amazing career so far. Congratulations on that. AI is, of course, all the rage right now, and it's hard to have a conversation that doesn't revolve around that topic. But from your perspective, what does AI adoption actually look like inside a mid-market company versus the enterprise narratives that we often hear about? I think it is very jarring in terms of what it's like for enterprise versus mid-market. It does very much feel like two different worlds. When I was at Morgan Stanley, I really saw how long it took to change things at scale. If we wanted a new tool, if we wanted to change a process, those were years-long planning.

6:34And so for some folks, they started the projects and they weren't even there to see it being executed or it going live. So I think bringing it back to what AI adoption looks like for the enterprise right now, I think a lot of it really is stuck in that pilot phase, that decision phase. Some of the recent studies that we've seen is that Deloitte found one in five companies has a mature governance model for autonomous agents. CFO Connect also found that 68 % of CFOs say they don't know where to start. So those are jarring stats in terms of where adoption or ultimately AI fluency is at. But when we look at the mid-market, I think not as much hype and noise around it.

7:24However, there's a lot of things happening. And I do think there's actually more movement, more teams going live with AI tooling. And I think part of that is really structural in nature. There's less analytics headcount to displace. So AI isn't replacing an analyst or a headcount that already exists. It's really kind of the first one a team would ever have. There's also less change management baggage. I don't say this lightly in terms of what mid-market and enterprise comparisons look like, but change management is a big thing. There isn't a committee that has to approve or defend how the old way of things has to be done.

8:05And so you're working with a much simpler system set or ecosystem from a tech stack perspective. So when you think about most mid-market, they're on your standard ERPs such as NetSuite, Sage, Dynamics, QuickBooks. They're not burdened by the customization, the multiple custom systems that need to be put in place. So while in practice, it does look small, well-defined workflows going live in weeks rather than months, we had one customer, for example, a school out in Texas that was able to centralize six and a half million in terms of spend and took approvals from a purchasing standpoint from over a week down to a day.

8:52And so I think to be fair with the enterprise, they're not behind so much as starting from a much harder place. Definitely a different starting position. So you're seeing different speeds in terms of adoption overall.

9:05Megan Weis: That's really interesting because I guess I would have thought enterprises would have it. They have more money to spend. It would be easier. But in a lot of ways, it's a lot harder for them to change direction of a gigantic ship. Exactly. So why do you think that many finance teams still struggle to separate genuinely useful AI applications from the hype? Or do they struggle at this point? I think it's a very real problem. I think it is also part of just who we are as finance folks. When you think about the typical finance or accounting individual, very much type A, very risk averse. So when finance has traditionally always been on the hook for accuracy, audibility, by nature, we're just careful and we should be.

9:57So I think we're just fundamentally wired that way. But there's a few reasons I think the hype is hard to see through. First off, vendors, I'm sure you're being sold to on a daily basis, they sell the model. They don't really sell the workflow. So while you may sit on a great demo, it isn't really a solution necessarily. But for me, I think before I really trust anything, I really want to see how the workflows are set up? What is the governance structure? What is the audit trail? And then what happens when the technology or the AI itself gets something wrong? I think that's the really telling part.

10:39The second is that I think we're measuring the wrong thing. A lot of it is flashy. Does it feel modern? I think if that's your mark on, is this the right tool? Should I trust it? you'll definitely get burnt in that sense. I do think we have to be more prescriptive with what tests we're using, and it should be very prescriptive in the sense and very measurable, like did it change a cycle time? Was there a specific compliance outcome? Or was there a decision that I can actually tie back to the tool itself? And then last but not least is really trust. Our company did a study in terms of AI readiness in finance, and we saw that about 37 % of organizations had said AI gave them more spend visibility, but only 30 % trusted the outputs.

11:31So when you look at that stat, you see that trust is always lagging adoption. And I think that's a gap. And that gap is real, and it's the state of things right now. But as we get deeper into the products, as there's more usage and testing, I think that gap starts to slowly close.

11:50Megan Weis: And as AI becomes more and more advanced, where do you believe that human judgment will still absolutely need to remain in the loop? For sure. I think at the end of the day, AI should augment your decision making, but it doesn't replace the accountability that still needs to exist. I think you can get things done a lot faster and at higher volumes. So when I think about it, it's really in tiers. So when you think about high stakes work, you definitely still want a human in the loop. And when you think about routine work inside things where you can build clear parameters, you want a human that's on the loop.

12:33So they're watching rather than doing. And all of that should be able to be traced back and auditable at the end of the day. I think where I wouldn't ever really take the human out is when you think about things like onboarding a new vendor or what off-policy spend looks like, when you're revising what a baseline for a budget is, when you're reviewing exceptions that are above certain threshold, anything that is critical judgment or judgment calls that are required, those are things I would always have human in the loop. I think where AI does really earn its place in terms of being effective is when you think about automation.

13:17So when you look at routine, repetitive, really well-structured work, any approvals that are kind of within your set policy, things like invoice coding or three-way matching, flagging exceptions, I think that's when it does its best work and it's able to narrow what your team needs to look at on a day-to-day basis. but it's not necessarily making the decision or core judgments for them. You'll still be in that loop. So I think for me, AI should augment decision making, but not replace that accountability. And so when something goes wrong, someone still has to be accountable and address the issue.

13:59Megan Weis: And I've read a lot about how AI is sort of replacing entry-level jobs and work. So do you Do you think we're at risk at all of automating away the roles that actually teach people about the business and how it operates? To be honest, I do think there is a risk here in the sense of I think it's a risk, but it's also, I think, an opportunity for many teams in terms of just understanding what their org structure should ultimately look like. I think a lot of the things, or I guess what I called grunt work or the things that taught me how to have that level of judgment I did at my entry level roles, right, or in the junior level roles, those are what is fundamentally being displaced by AI and tooling.

14:51Just if you have the right infrastructure or data layer in place, AI can be really effective in terms of getting those analysis to you quickly and with a lot more variations as well. So I don't think it's necessarily a problem that's been solved for at this juncture, but definitely I do think it does put at risk kind of some of those entry level roles where we've traditionally set those tasks to. But that being said, I think it gives you the opportunity to really evaluate your current work structure, right, in terms of where you have certain roles. I think leaner teams can now do much more versus before you would have had to allocate that capital to additional headcount.

15:36I think it's just a different way of doing work and learning. So we're working through that, I think, as an industry.

15:45Megan Weis: And you just mentioned like data needing to be in place, but what else needs to be in place before an investment in AI can create meaningful value? I think it is thinking about doing an audit or review of your current tech stack. One is from an infrastructure layer, what is the ecosystem you play in? What are the pain points you're actually trying to solve for? And what are the outcomes you're driving for? that needs to sit as number one and fundamental in terms of your evaluation. But also secondly, thinking about the people, the communication side. So putting on my operator hat, thinking about what change management needs to happen in terms of, hey, your responsibilities and scope are going to look very different.

16:36Your day-to-day used to be a lot of manual work or administrative work. Now it's on really high value or high stakes work in terms of judgment required. And so I think setting your team up for success in terms of having those conversations in advance and helping them be a part of that tool evaluation and helping them understand how a new workflow might fundamentally change the way they work. I think that is something that we don't really maybe talk about enough or have thought about enough, but also is, I think, a big blocker for a lot of finance teams in terms of hesitations to adopt technology.

17:20Megan Weis: And the tooling landscape changes incredibly fast these days. So how do you avoid constantly chasing the next platform or trend while still managing to stay innovative? For me, I think I try to keep it really simple. I use kind of three criterias when I'm looking at AI tooling. Does it really actually solve for a problem or is it just adding another layer of complexity and another bill that you're paying for? For me, I ask kind of, does it change a cycle time or a decision that I can ultimately measure? Does it work inside the system of record that I already have? So thinking about things like your CRM, thinking like other tools that you have in your tech stack, and does it make me adopt a new one and can it show its work?

18:13So if it's a yes on those three, I think it's real and I do deeper evaluation. If it's no, I think ultimately I might be buying hype and complexity.

18:23Megan Weis: And looking at the organization as a whole, how has AI changed or not changed the relationship between finance operations, procurement and the broader business? I think it's definitely changed the relationship between finance and the business more than the org chart itself. So I think it used to be that finance had to have the data and then all the other teams or everybody was had to come and ask for it. Generally, that can take days or weeks. And finance was often kind of seen as the bottleneck or gatekeepers of this. But now I think people can see that data in real time. And I think it's finance's job to really help shape the decisions being made with the data.

19:10So it's more about insight rather than gatekeeping. And I think when we think about procurement, procurement is near and dear to me in terms of a clearest example. It used to be the department of no. But now with budget guardrails right at the point of request, having conversations that happen before anyone actually commits, it's not a conversation that happens weeks later when the bill lands and when we're making the payment and a reconciliation is happening. So I think the speed is changing the tone in which where, hey, if approvals used to take 50 hours, now it's taking seven hours. Finance really stops being that bottleneck or the workaround people see.

19:57It's a relationship change and not necessarily a process change.

20:02Megan Weis: And you might have just answered this question, but you mentioned it being the office of no. I've heard it referred to the role of a CF no. So how can finance leaders use AI and automation to improve discipline without slowing down their teams or the business? There always needs to be checkpoints. Sometimes I describe myself to my team as kind of that speed bump in a road in terms of before we move too fast on a decision, let's do the speed bump check. But what I would put is like bureaucracy is when you design a checkpoint for that worst 5 % of cases and then you apply it to every case. What we want to drive for is discipline.

20:46So when your system is able to enforce policy automatically and then only the exceptions then come to you. So in practice, that means putting budget guardrails at that point of request, soft stops and hard stops built into that intake process. And approvals being automatically routed by department, threshold, and vendor risk. And that's what we have been able to build in terms of our guided intake product. So I think at the end of the day, accountability doesn't necessarily go away, but it does move the team and people from manually having to check things to policy. And then so you're seeing like that same level of accountability, but definitely a lot less friction.

21:34Megan Weis: And last question, but as you look ahead, how do you see the role of CFO evolving over the next two to three years? Because there's no way we can see out past that point. Five to ten years is long. Yeah. Maybe even kind of the next two to three. I know. Yeah, for sure. Yeah, I think it is going to be a lot of learning. I think where we've traditionally been that record keeper, right, in terms of reporting on things that have occurred, we kind of rebrand and move towards more of a value architect, where we help kind of design and determine how value gets created in the organization. So practically speaking, I would say that means the CFO role is in the room earlier, right?

22:24When product is being launched, when a go-to-market plan is being built, and not really after the fact when decisions and dollars have already been committed or moved. I think also when you think about like real allocation or cash management, it gets to become a lot more continuous. The traditional way of annual budgets won't necessarily disappear because boards at the end of the day would still want an anchor, but it stops being really like the only time we would make those decisions. So you're continuously rebalancing, reallocating as you go, provided that you have those clear guardrails. And then I think really the CFO does become a systems designer, building out that finance ops that's able to continuously run instead of just being kind of burdened by the one -offs or the month-end cycles.

23:22and we become that storyteller as much as a steward because numbers on their own don't really move or motivate people or the business. And I think the pipeline changes. So I think in terms of when we talked about what does the next generation looks like, the next generation of CFOs will look less probably our traditional set of controllers who moved up and probably more like operators. who are able to manage the numbers, but also leverage the broader skill set.

23:58Megan Weis: Amy, thank you so much for taking the time to be here with us today and share your experience and knowledge. Thank you so much. I really appreciate you having me. And to all of our listeners, please tune in next week. And until then, take care. Thank you for listening to this episode of CFO Weekly. We hope you picked up some useful insights and strategies to help move your business forward. If you enjoyed the conversation, be sure to subscribe on your favorite podcast platform so you never miss an episode. And if you found it valuable, we'd love for you to leave a review and share it with your network.

24:33Megan Weis: It helps us bring even more great content to finance leaders like you. For more episodes, resources, and to see how Persona can support your accounting and finance needs, visit persona.com. See you next week.