Introduction

If you are a CFO, FP&A leader, or finance professional trying to cut through the noise around AI and understand what it actually means for the way you work, this conversation is worth your time.

In this edition of our Fireside Chat series, "The Dream AI Agent for a CFO," we sat down with Brian Kalish, Principal and Founder of Kalish Consulting, a seasoned FP&A expert and CFO who has led finance teams across multiple decades. Moderated by Niyati Chhaya, Co-founder and VP of AI/ML at Hyperbots, the session was grounded, specific, and honest about where AI in finance actually stands today and where it is heading.

Brian came into this conversation fresh from Southeast Asia, having just returned from Brunei, Indonesia, Malaysia, and Singapore. His opening observation set the tone for everything that followed: the challenges the office of finance faces are not regional. They are universal. Whether the conversation happens in Brunei, Dubai, or London, finance leaders are wrestling with the same problems, the same confusion, and the same desire to find something that actually works.

Key Takeaways

• Most finance leaders cannot tell RPA from Agentic AI. That gap in understanding is stalling adoption more than any missing technology.

• Finance teams spend 80% of their time getting data ready and 20% using it. That ratio is the problem AI exists to solve.

• Applying compliance-level accuracy standards to forecasting is how finance teams over-caution themselves out of useful AI tools.

• Machine learning on a 150,000-customer dataset revealed 40,000 unprofitable accounts. Without that, tariff strategy is guesswork.

• IT will not solve this for finance. The office of finance has to identify the tools, evaluate them, and drive adoption itself.

• Gartner predicts 15% of finance offices will be autonomous by 2030. The gap between now and then starts with getting a 45-day forecast down to five.

Summary of the Conversation

Meet Brian Kalish

Niyati: Today I am happy to have Brian with us. Brian is a seasoned FP&A expert and CFO who has been leading different finance teams over multiple decades. Welcome Brian, anything you want to kick off with?

Brian: Thank you for the invitation, wonderful to be back. The only thing I would share is that I recently just came back from Southeast Asia. I was in Brunei, Indonesia, Malaysia, and Singapore. And again, what is fascinating to me is just the universal challenges that the office of finance is facing. Whether it is in Brunei, which I had never been to before, or whether we are in Dubai or whether we are in London, what we are talking about today is truly universal. It is not a nationalistic or a regional challenge that people have, it is truly universal.

Finance Leaders Want Practical AI Today, Not a Promise About 2030

Niyati: What are you hearing from all the folks that you have met so far?

Brian: There is a lot of buzz about AI and technology. And I think there is a gap, in all honesty, between understanding, well, two things. One is, I think people need to understand what we are talking about when we are talking about things like RPA versus GenAI versus Agentic AI. And there is a little bit of it can do everything. And so I think there is kind of like the overpromise. And then I would say the third point is what they are really keen on is how do I bring, what is tangible? What can I get done today? And so they are hearing on one side that AI is going to take over the world. On the other hand, they are like, how do I apply that today? So I think one of the critical roles that we can play is help people understand what is the art of the possible. And certainly still being aspirational to where we are heading, because we certainly can talk about Agentic AI. To reference Gartner, last year nothing was done in Agentic AI. And by 2030, 15% of the office of finance will be autonomous. And so it is kind of hard to grasp with that kind of change. I think the other piece too is just the velocity, right? Things are changing so fast. And for the office of finance specifically, because we are small-c conservatives, how can we bring it in manageably? The term I use is future-proofing the office of finance. The idea of how do we build something today that not only solves the problem for today, but sets me up to thrive going forward.

Why Financial Data Is Too Sensitive to Hand to an AI Without a Moral Compass

Niyati: It is not only security, there is also a notion of responsible AI. We are talking about a dream AI agent, but how do I make sure that agent, even if it is very useful, is not doing wrong things?

Brian: Going rogue.

Niyati: So going rogue. It becomes an adversary. What do I do there? So it is very, very critical that that does not happen.

Brian: Right. Just because, not ever knocking any other part of the business, but the financial information can be very, very sensitive. And the idea of just opening that up. You have a human doing it today. AI does not have a moral compass, right? Sometimes it may not necessarily understand the sensitivity of the data that it is working with. From the CFO's perspective, we are trying to overcome that because obviously it can be done, but it is just that, hey, I know if Bob sees that information, he knows, oh wait, I should not be looking at that. Whereas how does an agent learn that? And part of it is, kind of being a doctor, do no harm, right? I might be behind the curve from a technology standpoint, but that is better than hurting the patient.

What CFOs Actually Do Every Day: Cash, Risk, Compliance, and Strategy

Niyati: What are your daily tasks as a finance leader or a CFO?

Brian: Very broadly, you are never going to get away from just financial reports, right? What is going on in your core financial accounting, what is happening with the P&L and on the balance sheet, income statement, but then also understanding what is going on with cash flow. And of course we are always living in uncertain times, but we are in really uncertain times right now. The term that I use is VUCA, volatility, uncertainty, complexity, and ambiguity. And I think it is safe to say we are at a high right now on VUCA. And so just the criticality of cash, of working capital, of liquidity. Having built out ideally a good dashboard with my KPIs to see where things are trending, what is going on with the budget, what is happening with our financial plans. The other two areas once you get up to the CFO level are risk assessment and compliance. I look at it from an R&O perspective, risk and opportunity. What are the risks that are out there, but what are some new opportunities? And then we are just never going to get away from compliance. That has become so much more critical over the last 20 years. That is kind of how the day just starts. What are any burning platforms we have? And how does that then flow up into the strategy of what we are trying to do?

The CFO Should Not Be in Excel 80% of Finance Time Goes to the Wrong Work

Niyati: How often do you end up making Excel sheets or consuming them to get insights? Or is that mostly delegated out?

Brian: That is mostly delegated out. From truly sitting at the CFO level, that is pretty much delegated out. I would actually go the other way in the sense that you should not have your CFO involved with Excel spreadsheets. And I have worked with people, and we love it, right? Those of us of a certain vintage, that is what we grew up on. And sometimes I have worked with other peers where it is really hard to take that away from them. They want to do the work because they love it. It is trying to explain to them, it is just not a good allocation of resources. The CFO can be the best Excel person that you have on your team, but that is not how you should be using their resources. 80% of my people's time is on data acquisition, verification, reconciliation, and 20% is on analysis. I come in every day and I am trying to figure out how to flip that. I want my people to focus on the whys, not the whats.

Why Finance Professionals Who Use AI Will Outperform Those Who Do Not

Niyati: Would you want AI to do the whys?

Brian: Not fully yet. It really depends. It is fascinating what is out there, right? Because there is just this gap, and I attribute it more to education. The office of finance is already working 30 hours a day, 10 days a week, 75 weeks a year. And the challenge is, when you are so focused on just getting through the day, how much time can you allocate to improving the situation? I need to hear from my treasurer. I need to hear from my controller. I need to hear from the folks that are involved with AP and AR. When we do these things, are there better ways of doing it? There is just this chasm between understanding what is possible with technology and where we can be. I am someone who believes so much can be, I would say AI'd away for lack of a better word, but it is finding that right combination between the humans and the AI. AI is not going to replace people. People that use AI are going to replace people that do not use AI. It is just a skill that you are going to have to have. But the onus is so much now on finance to seek out that technology. IT is so busy with so many other things. It needs to be driven from the office of finance.

Knowing Which 40,000 Customers Are Unprofitable Changes Everything About Pricing Strategy

Niyati: How often are you involved in business model discussions, say with the CEO, identifying the pricing, identifying the target market?

Brian: Oh yes. And increasingly so, because we have the skills and the tools that they do not necessarily have. The elephant in the room right now is tariffs. And understanding what has happened in an incredibly short period of time is understanding what the true cost and value of your customers are. To give you an example. A company in Chicago, let us say they have 150,000 customers. Well, the old Pareto 80-20 rule will tell you 20% of those customers are generating 80% of your profit. We actually got in there and looked at the data, leveraging machine learning, and it came out to be about 3,000 customers generating 80% of the profit. Around 107,000 producing that last 20%. And about 40,000 that are unprofitable. So think about understanding that, and then adding in a situation where you have a tariff variable. There are 3,000 customers you do not want to pass the full cost of that tariff along to, because they are so valuable to you. But the 40,000 that are not even profitable, pass that along. Being able to give that insight to the business. It is not just revenue, it is understanding what the cost is. And at the end of the day, what we really care about is profit. The cadence of talking to other members in the C-suite is only increasing.

The CFO's Role Is to Ask the Right Three Questions, Not Produce Every Report

Niyati: How much of the clerical work that goes into talking to CXOs or investors is actually owned by the CFO, or is that also delegated?

Brian: I would say it is topic driven, right? Part of it is just size of organization. From a regulatory, tax, and accounting perspective, that is coming from my leaders in those areas. When you get more to data analytics, that is more pulling the team together through FP&A, which is the path I came up. I am in the role of asking the questions, anticipating the questions, understanding what the business, what the CEO, what my peers in the C-suite want to know. What I try to do with my partners is understand: what are three things you would like to know today that you do not know, that you think could really impact the organization? And that is what I am focused on. Can we get to that data? Part of the challenge is the data is in many, many different places. That is where Agentic AI comes in. Agentic AI is fantastic for pulling in information from very disparate sources. What is the value of a monthly forecast that takes me 45 days to do? Getting that 45-day process down to five. I have to leverage my people and technology to get there because the world is just moving too fast.

Why a 4% Forecast Landing at 3.75% Is Good Enough, and When It Is Not

Niyati: How do you think about the difference between the accuracy requirements of accounting versus analytics?

Brian: If you take something like tax, if you take accounting, they have to be right. It is binary, it is either right or it is wrong. And if you are wrong, you might go to jail. Whereas when you are talking about data analytics, all forecasts are wrong. It is the old George Box line, all models are wrong, some are useful. So it is understanding what is good. And what we are trying to do is not be perfect. It is to be better. At the end of the day, if we are the best in our industry, that does not mean you are perfect, that just means you are the best. If what I need to know is whether we are going to grow by 4%, and it comes in at 3.75%, we should understand that but not spend an incredible amount of time on it. That is pretty good accuracy for a forecast. Are they going to go 20 and 0? No. But if we go 15 and 5 and win the championship, that is what we are trying to do here.

Why CFOs Trust Treasury Forecasts Over FP&A, and What AI Should Do With That Data

Niyati: What are the documents you look at daily? Where are they stored, and how do you generate the insights you need?

Brian: It is looking at, for the most part, what is our plan, what is our forecast, and the variance from what we expect. And that is being generated by many different parts. When we talk about liquidity and cash position, that is coming out of Treasury. I have a Treasury background and I will tell you, the cash forecast from Treasury is my preference over the FP&A one, because Treasury says, this is what it is. I am a believer that should is a bad word. This is what it should be is not helpful. This is what it is, is what I need. And then understanding why. Are we at a burn rate that makes sense? If I am looking at the data analytics teams, when they start detecting trends, that is what the game is all about now. Can you detect trends earlier? Is it an anomaly? Why did it happen? We are spending much more time on diagnostic and predictive analytics than descriptive. We cannot just spend our time looking backwards. The opportunity I see is tremendous, in developing co-pilots that really work with different parts of the organization to move things off the human's plate that are not bringing a lot of value add. Most activities can be around 80% automated, but the human is the 20%. And that 20% is what is going to impact 80% of your profits. The other 80% is just straight through.

How Hyperbots Is Helping CFOs Get There

What Brian described throughout this conversation, flipping the 80/20 ratio, surfacing anomalies before they become problems, connecting fragmented data in near real time, and freeing the finance team for strategic work, is exactly what Hyperbots is built to deliver.

Hyperbots' purpose-built AI co-pilots for finance operations address the specific processes where manual effort currently dominates. On invoice processing, Hyperbots delivers 99.8% extraction accuracy and 80% straight-through processing, eliminating the manual work that consumes AP team capacity. Reconciliation costs are reduced by 80%. On collections, Hyperbots brings a 70% cost reduction and a 40% improvement in DSO, addressing the exact cash flow visibility problem Brian described as a daily priority. Cash flow monitoring shifts from periodic to continuous, with alerts surfacing before problems arrive rather than after. Unapplied cash is reduced by 90%.

Every action Hyperbots takes is logged in a complete audit trail. Human oversight is embedded at every critical decision point. The system integrates with existing ERP infrastructure without requiring a full replacement, deploys in days rather than months, and is designed to earn trust through demonstrated accuracy before being extended to higher-stakes processes.

The CFO who finally has 80% of their team's time back for analysis and strategy is not a future state. It is what Hyperbots is delivering today.

See it in action with a demo or start your free trial today.