Welcome to another episode of Agent CFO. And today we are with Dave Sackett, a great CFO, great finance leader who worked in many roles. He is one person who has been talking about AI since 10 years, AI in finance. He saw what is coming even before ChatGPT was in picture. And he has been talking about it and how things are going to change, what are the processes that will be first automated using AI for really long time. I have privilege of knowing him from last few years, and I have blast discussing all these insights. Welcome, Dave.

Yeah, thank you, Ashok.

What are you doing these days?

I'm a VP of Finance for Persimmon Technologies. We make vacuum robots for the semiconductor industry. So pretty exciting technology and time to be in semiconductors.

You are in high demand right now. That's awesome. Do you see this big push for AI that has impacted your sales and the demand?

Absolutely, yeah, AI chips specifically are driving the market. Huge investment, huge production runs, big demands on semiconductor companies right now resulting from AI and the explosion of how popular it is.

How are you handling it? More Excel files?

Yeah, we do use a lot of Excel files. We had a new ERP system put in last year to help scale us and get my team up to speed before we really start chugging away. Yeah, we've got bookings are good. We've got a plan to really grow in the future. So the idea now to scale in the future appropriately is kind of my role.

And I think it's exciting times as well. It will be a little scary for finance. When you're in a stable business, it's totally different how you can manage everything in finance. But for a business like yours right now, with the high demand coming in and you don't know where it is going to take you. It's really difficult to manage the forecasting, the budgeting part, the long range planning.

That's right. Yeah, you've got market forces, you've got new product forces, you've got new customers coming on, new processes coming on. So everything's new and everything's growing. So you don't really know what to expect. You just have to be ready.

It's a good problem to have. We spoke about this for a really long time. What is one of the dangerous phrases in finance? What do you hear?

Yeah, the most dangerous phrase is when people get lulled into not really thinking about the future and just if you ask someone why things are done that way, they're very content to say, well, we've always done it this way. So let's just keep doing it this way. To me, that's so destructive of a way to think in terms of adopting new technology, new processes, really self-examining how you're working.

That's right. It's like we have done this the same way all this while, right? Breaking it is not about the individual motivation or smartness or intelligence. I think it is breaking into the process of our work itself.

Right, yeah, how can you do your work more efficiently? How can you have low value tasks not matter?

And it's also, I make decks and presentations sometimes. When I conceptualize them and I design, I actually miss the overview of what we are making. But that happens with finance work as well, right? You are in the weeds of it, dealing with all the details, then you miss the bigger part of it.

That's right. Especially when you're scrambling trying to pull data together for a report and then you have all the details but then what does the report say is the trend? What is the meaning of it all? That can get lost for sure.

Is there a better way to manage this — being in the details at the same time having that 30,000 feet view of how things are flowing?

Yeah, I would say automation. The more you can automate and speed up your data, the more time you have to analyze.

That's a good way of seeing it. If we can automate a lot of the details, then we don't have to deal with it. Now we can deal with an abstracted level of processes. Abstraction is one of my favorite topics these days. At Chatfin, we make people build finance applications, whatever they need. With the right form of abstractions, as a finance user, you don't have to give all the details. For example, you say “I need balances of my fixed assets accounts” — you don't have to mention how to get that data.

Yeah, it's designing a process that has help. So you're not doing everything. You're only doing things that AI can't do if you're going to automate processes. Pulling in agents to help out with forecasting or to help pull together reports because they're goal-centered and they can act autonomously. To me, that's amazing, where they can just go get the data based on what they know.

What's the most expensive assumption that is made in finance teams?

I think when it comes to technology, for example, like going to an ERP system, that you need to spend a lot of money to get an ERP system that's going to work the best. And I think that's just an assumption. If you're going to spend all this money, they must have all this functionality, they must have done all this testing. So that should be the way to go.

And also people underestimate the effort that goes in an ERP implementation.

It can be a huge project that needs lots of planning, lots of control, lots of testing, lots of feedback from people as you go.

And all the manual processes that are built earlier, usually you can't lift and shift to the new ERP. That becomes a nightmare.

People put their old process onto a new ERP system and it wasn't designed that way. So I tell people, speak the language of the ERP system. They designed it to work a certain way. Learn that way. See if it can be efficient in the ERP system.

With all these new ERPs coming in and the talk about making it AI first — what's your perspective? Do you think ERPs, which are systems of record, should combine with AI, or should AI sit outside?

My bias is to have your ERP system be kind of standalone as that system of record. But certainly putting workflows and AI and everything else on top of that would be my preference.

One clear pattern we see is point solutions. We used to have a lot of point solutions in finance and they worked well. But now with this AI era, I don't think we need them anymore. You don't need a separate close management software, a special reporting tool, or a special working capital management tool. All you need is a good system of action that gives end users the ability to customize it and build these small point solutions with an AI layer.

Yep, that's certainly a solution.

Looking back at your career, is there something you would approach differently today?

Knowing where we are today, investing more in people, investing more in really rolling out someone's future in finance and accounting in terms of what's their path forward. Doing more of that, really trying to grow with people, give them skin in the game to stay at the company and not have this constant turnover. Really show them a future of where they can go if they're good performers.

Many leaders say empathy is a great characteristic for a finance leader. You have your eyes set on the goal, but achieving it you need people to march towards it. You can do it with a whip, or you can do it with empathy — probably a combination of two.

I agree. You need a good combination.

Where do you think the future of ERP is going to go?

I do think there'll be a merger of ERP and AI. Every piece of software will have AI attached to it, whether it's built in or sits on top. There's going to be some connection that utilizes AI.

Do you think the legacy ERP companies will survive?

I do. I think they'll adapt. They'll have clean data, tables that are very AI friendly for AI to manipulate and pull data out of. Companies that aren't doing that could be in trouble. But if you know that people are going to use APIs to dig into your databases, making that as easy as possible is the way of the future.

There are tens of other ERPs for specific industries — manufacturing, service, restaurant. They need all those nuances built into their ERP process. It's not a question about ERP versus AI ERP. It's how can your ERP bake in all the details for that specific industry, and on top of it, add an AI layer or connect to other tools.

Especially when you have specific industries, whether it's steel, woodworking — they have very unique needs. If your ERP can address those and make things more efficient for them, that's going to give them success when selling into an industry they support.

You've done extensive research on AI tools. Is there a feature or capability that really impressed you?

Handwriting — to be able to identify handwriting and convert that into English and into numbers. That amazes me that that's even possible.

We actually deal with that. In some companies they still use handwritten documents. A lot of people think it's just OCR, but you need much more than OCR. What if you can't even read it as a human? O and 0 are always the problem. There should be multiple validations. It's one of the toughest problems to solve.

And one problem is Excel files with merged cells. Merged cells is kind of a blasphemy in finance.

It used to be my mantra — I would never use merge cells. I would use center across collection, so many different tricks. But ironically, I find myself merging cells all the time now.

With AI, merged cells is a really big problem because it gets confused between which column and which row data belongs to. A lot of people say they have AI in Excel, but the real question is how do you handle all these exceptions?

Directionally, how do you think AI is going to change the finance process in the future?

I think people are going to have assistants. There'll be AI assistants that you can say: run me this report, do this variance analysis, get all this data, put it together, give me English in terms of what's happening. Give it your first go. I think that kind of work is going to take time off human finance leaders' plate.