Niv Yaar: Hello everyone, great to have you here with us. Hello, Rachel.
Rachel Joel: Hi.
Niv: Today's topic of the webinar is "Closing the Gap with AI: Controllers' Role in Real-Time Finance," and I'll just add: the controller's role in the era of intelligent finance. I think you'll see that that's part of the focus today. Rachel, let's just start with short introductions. I can start. My name is Niv, I'm the Chief Business Officer of Panax, one of the co-founders of the company. I come from a finance and business background, so I'm a CPA. I spent some time at EY in financial consulting, moved to the private equity world, was both an investor and an executive in a portfolio company, and started Panax three and a half years ago with two co-founders. Since then, I've been growing the business and am in charge of all the go-to-market, customer journeys, partnerships, and all the commercial side of the business. Rachel?
Rachel: Okay, mine's a little bit less impressive, but I'm Rachel from Melbourne, Australia. I started my career at PWC as a CA and then moved on to another fintech called Airwallex. I've been here at Panax for over two years now, working across product and customer, and helping navigate the direction of the product.
Niv: Awesome. You're too humble, but as one of the first employees at Panax, I think you have one of the deepest understandings of the bits and bytes of the system, how it works, and the value we're providing to customers.
All right, so we'll kick off by describing a bit the role of the controller and how it has been evolving over the last decade. I think we're all feeling the shift of expectation for the controller to not only play a technical role but also to be a strategic partner to the CFO, to the business, to the management, and to impact how decisions are being taken in the organization.
We divided it into three main legs that the controller is overseeing. The first one is accounting—obviously the more technical part of keeping the books updated, aligning the numbers with the accounting standards, and making sure that the books reflect the business. The second one is more about the day-to-day operations. Depending on the size of the business, usually what we see is that controllers are also in charge of collections, payments, and sometimes even cash management, at least until a treasurer steps in. Even then, we still see that controllers have an impact or an interaction with the treasury team around that. The last one—and I think that's also the opportunity of how to leverage AI—is to really play a strategic role, to be a partner for the CFO and leadership, and to drive business decisions.
We just added a quote we saw in an EY blog that describes the shift between technical accounting and strategic partnership, highlighting that controllers shouldn't only be a part of that change but also need to be the driver of the transformation and lead it.
Now, what are the key challenges of that change? Basically, we see three main challenges. First of all, the data is scattered across many systems and platforms, and it is unstandardized. That's part of the challenge of capturing everything in one place and creating a database that can be used across many processes. The second one is limited resources. Finance teams, by nature, are lean. Professional personnel is hard to hire, and we see that most of the time is spent on manual processes—collecting data, crunching numbers, tagging information. By the time you get to the level of creating insights, it's already too late and outdated.
The last one is insufficient capabilities or insufficient automation. We see that in many companies the tech stack is quite outdated, and even with a new stack, it's challenging to get real automation in place. It's hard to make sure that the data is aligned and that the day-to-day can be built on top of that. With relation to AI, most systems don't use real finance context, and without that context, it's really hard to gain the value of insights to drive decisions or to standardize data. We added a survey from Cherry Bekaert, a business consultancy firm, which asked CFOs from mid-market companies about their challenges; 55% mentioned that data accuracy and consistency is a major challenge.
Now let's talk about the era of AI and how it changes the game. First of all, let's define AI. I guess it won't be very new for you, but just to align: AI refers to systems that can learn, reason, and drive smarter decisions using data and algorithms, in a sense replacing the human factor in the process.
Getting back to the finance day-to-day—and then we'll apply the AI to it—what we see today is that finance teams and day-to-day operations are managed mainly with two data sources. One is the banks, where the cash data is. The other one is the ERP, where the accounting data lives. Basically, the two data sets are not synced, and it takes a great effort for finance teams to standardize the data, connect the different systems, reconcile transactions, and only after that, gain insights and analyze everything. That's part of the data challenge. That's reality today: two systems of records, very ancient, not very insightful, and the connection between the two is hard to manage.
What do we want to achieve, or what's the end goal here? First, we want a complete database. This has two aspects. First, we want to cover all data sources so the data can be complete and cover a full 360-degree view, whether the company has many banks, platforms, or multiple ERPs. In the end, everything needs to be integrated into one place. Second is the way the data flows and the infrastructure is built. You want the data to flow on an ongoing basis to ensure it is up to date and reaches the relevant places. If you manage AR on an AR system, you want the invoices and the bank data to flow into it. Same for the treasury aspect; if your challenge is to manage cash and you have a treasury system in place, you need to make sure all bank and bank-like platforms flow data into the system on an ongoing basis.
The second aspect of the data is creating standardization across the different data sets. That's definitely a challenging one because of timing differences between accounting and cash, but also because of different formats, even among different banks, and when combining accounting data with cash data. It needs to be standardized and synced.
Although it is challenging, the value of getting there is huge because once you have that complete database and standardized data, you can drive many processes automatically and insightfully. For example, you can reconcile transactions automatically, optimize cash, create streamlined and automated forecasting, automate reporting, and detect anomalies. You'll be able to see how the AI works much better and trust it as different agents can be deployed to complete tasks and do the heavy lifting for you.
I want to get into the more practical discussion of how AI is used in finance teams and in cash operations specifically. First of all, AI is not a new practice. It has been used for many years, mainly in the form of machine learning (ML), which means smart algorithms can learn out of past historical trends and predict the future based on that. A practical example of that would be anomaly detection; if the model detects any anomaly, it can raise a flag. Additionally, it can be used for prediction based on historical patterns or automatic tagging of transactions based on past tagging.
The main innovation introduced in the last few years are obviously the language models (LLMs) that we all hear about and use daily. These models don't just calculate; they understand context and generate content. The practical use changes to places like giving a prompt to the model to see different scenarios for planning. Or, when looking at reconciliations, it can now use context from different data sources to automate the process and suggest matches to the user. Another practical example is tagging transactions. With an LLM, the model can learn out of the context of bank descriptions, journal entries, different notes in the ERP, and external sources to offer different categorization rules and make life much easier for the user.
Let's get into how Panax uses AI practically today. But before handing it over to Rachel to walk us through the platform, I want to give a few words about Panax. Who are we? Panax is an AI-native cash operations platform built for lean finance teams with complex treasury needs—whether that means different banks, platforms, currencies, or entities, the common denominator is complexity. The company was founded by finance professionals, so we really speak the language and understand the customer's needs.
Today, we touch on four main aspects. The first one is treasury: we streamline all banking data into one place, giving full control and visibility over different pockets, optimizing cash, and forecasting the future. The second is accounting: we sync between the two data sets with reconciliation and posting journal entries, saving tons of manual hours. The third is finance operations: we have an Accounts Receivable (AR) module that gives insights based on actual customer collection behaviors and an aging flow to manage day-to-day collections. The last one is payment operations: customers can execute cross-border payments with FX and utilize investment options through the platform.
So, how do we use AI? We built Panax with an AI-first, AI-native approach to the structure of the data so a lot can be built on top of it. Here is the list of agents in the system or on the roadmap. First, on the connectivity, visibility, and cash flow side, we have agents that monitor data and create insights, like automatically identifying cash in transit.
On the forecasting side, trends based on actual collections and historical data are used to predict the future. We are also introducing a liquidity optimization agent. When you have different pockets of cash and need to be capital efficient, this agent will recommend and change the optimal capital allocation across the different accounts based on actuals.
On the operations and control side, we have a reconciliation agent and an AR agent that reminds you of overdues, creates emails, and shows insights into potential late collections. An AP agent will be introduced next year. On the risk management side, we are creating agents to monitor bank and currency exposures proactively so the customer can keep risks mitigated. I'll move now to Rachel to walk through the system and show practical examples. Rachel, the stage is yours.
Rachel: Thanks Niv, and you're welcome to join in whenever. I will share my screen. I'm assuming you can see my screen now.
The first thing you'll jump into as you land on Panax is the dashboard, which gives you an overview of what's going on, the current status, balances by various breakdowns, inflows, outflows, and more. For the purpose of Niv's intro, we'll try to focus on the connection between the banks and the ERP data sources, why it's crucial to keep these connected, and how Panax facilitates this process.
If I jump quickly to the cash position, you'll see the aggregation of all balances from your different bank accounts, split by various types. If you're a global company with over 50 to 100 bank accounts, logging in every day to understand your cash position is really difficult. So that first layer of data is centralizing it and making sure you have an accurate base point to make cash decisions.
From the other side, Panax connects to various ERPs—even those known to be difficult to connect to. The first way Panax helps in this reconciliation between the bank and the ERP is via the bank data feed. Instead of logging into banks, downloading statements, and uploading them to your ERP, Panax automatically syncs that data straight to your ERP to prepare you for the reconciliation process. You can briefly see the number of transactions synced per day, the amount of banks covered, and the accounts synced.
Once everything is reconciled in the ERP, the next piece of the puzzle is categorizing cash transactions for reporting and forecasting. The other way Panax syncs with the ERP is through an AI algorithm that identifies matches between objects in the ERP (like a bill payment, invoice payment, or journal entry) and bank transactions. For example, you can see an incoming transaction of $15,097 and the matched invoice payment from the ERP, showing the GL path and associated customer. You can also trace the mapping between the customer in the ERP and the cash category in Panax.
Another way Panax categorizes transactions is via rules. We have an AI model that identifies common patterns in transactions based on descriptions, amounts, dates, or transaction types. Panax will often suggest the most relevant category based on its AI model.
Niv: Maybe you can show the AI-created rules on the smart categorization banner?
Rachel: Sure. These are Panax's AI-based suggested rules. The model determines recurring descriptions and patterns and provides suggestions for rules you can create instead of creating them yourself. All of this categorization is for the purpose of cash reporting and cash forecasting. For instance, the cash bridge report shows your starting balance, closing balance, and any movement in between split by customized categories.
Panax also helps speed up the reconciliation process, particularly around closing AR. The AI model identifies incoming bank transactions tagged as collections and matches them against open invoices in the ERP. This is a huge time saver. Automatically, matches are posted to the ERP with the click of a button. We have global companies with over 20 entities where the AR team previously spent up to 8 hours a day reconciling incoming payments to invoices, and with Panax's cash application, they've reduced that by 75 to 80%—down to 1 to 2 hours.
The model runs using a number of factors. Sometimes it's obvious, like finding the customer name in the transaction description, but it also handles complex examples where there's no invoice number or customer name yet a match is still found. We can also find matches on multiple invoices with one incoming payment, or close payments involving fee differences or discounts.
Once everything is matched, it can be synced to the ERP with the click of a button. If you're more data-conscious, you can export and upload it directly to your ERP as well.
Niv: Just to add, if you have an exotic ERP that isn't easily connectable, we can modify the export file format so you can use it to upload, or we can auto-export it directly to the ERP in different ways.
Rachel: Yes, good point. All these things mean you can close the books faster, understand your current position, and forecast future collections more accurately.
Jumping to our accounts receivable tab, we dive into the details of how customers typically pay against payment terms. You can see a graph showing average days to collect against average payment terms, cut by individual customers or entities. This is where the system learns typical behavioral patterns to provide granular insight into exactly when you plan to receive collections, rather than simply relying on a number given in the invoice due terms. We also have an aging report showing overdue invoices and everything owed to you.
Niv: Getting back to the collection report for a second—I wanted to add that we see different use cases where companies calculate DSO differently. Building the DSO bottom-up here, based on actual behavior, is much more insightful than doing a high-level balance sheet calculation. The consequence is you can rely on much more accurate data for your projections.
Rachel: I think a good visual for that can be seen in the forecast. When we forecast collection categories, having that insight from the ERP makes the forecast much more accurate. You can choose to forecast based on contractual terms (the invoice due date), or you can choose to forecast based on customer collection trends or entity collection trends. Many customers tell us that using expected collection patterns makes their collections much more accurate. If we preview the data using customer collection trends, the numbers shift in accordance with expected payment patterns. You can also manually change individual invoice dates if you have inside knowledge of a delay.
Niv: Just to add, that's using the ML side to understand historical patterns. For completeness, you can use our AI engine to show data trends as a baseline, which you can then step in and adjust or override.
Rachel: That's a good point. Another way we push these insights is via proactive alerts on the dashboard. It will notify you if expected collections are lower or higher than contractual collections, and point out recommended actions—like identifying high-risk customers or overdue high-value accounts. It gives you a proactive way to stay on top of collections.
Finally, a new and exciting feature is Panax's own ChatGPT-style interface where you can ask basically anything about your data in the platform. For instance, "How are we tracking on collections?" It gives you an answer showing overall performance, areas to pay attention to, positive trends, and recommendations. That’s the end of the demo today. I'll hand back to you, Niv.
Niv: Thanks Rachel. To take a step back, we wanted to show how Panax closes the loop and manages the cycles of data. We automatically tag transactions with AI, reconcile them into the ERP, show actual collection trends, and flow it all into the forecast so you rely on accurate data.
Just to wrap up with a few real-life use cases: Go Global has more than 50 banks, and we save them 50 hours a week of manual data collection. Diligent uses cash application to save over 75% of manual reconciliation time. Time Payment reduced their forecast discrepancy to 1%. Optimum saw capital optimization of more than $5 million translated into a pure ROI yield of over 200k a year.
What makes us the best solution in the market? We automate workflows to achieve autonomous finance for real. The system is built by finance professionals who understand the customer's pains, and we offer a very quick time to value, holding the customer's hand throughout onboarding. Thank you everyone for being with us today, and thank you Rachel for doing the great demo. We’re happy to show you the power of Panax and the automation that many finance teams already leverage.
.webp)
