Gurkan Salk, Microsoft General Manager of Copilot for Finance, and Leigh Pepper, Aptitude’s Chief Product Officer, discuss the recent launch of Microsoft Copilot for Finance, the product vision and how it helps drive an Autonomous Finance function.
Mark Aubin: Gurkan, first off, congratulations on the successful launch of Copilot for Finance. How did you evaluate and prioritize capabilities and use cases for the initial launch?
Gurkan Salk: Thank you, and that’s a good question, Mark. From a product development perspective, we wanted to prioritize the most impactful use cases first – those that would help finance professionals be much more efficient and productive and get the right insights. We then picked use cases that had the broadest reach – horizontal use cases that most finance professionals use, like data reconciliation.
Traditionally, we would have asked customers what the systems should do. Now, we recognize the need to change that culture and ask, what can the systems do? That has changed our product development cycle. Throughout this whole process we have worked very closely with our early adopter customers, including in our own Microsoft finance department.
Mark Aubin: I was just thinking about how this process represents a shift in product development and product management. It’s no longer just listening to existing customer requirements and then designing a solution. Now, you’re actually helping them uncover new requirements because you’re at the forefront of technology. How did you work to shift your product team’s mindset from listening and then figuring out a problem, to finding new answers to problems that people might not be sure they have yet?
Gurkan Salk: Just like we’re shifting the culture of the finance department, our product team also went through a culture shift. Developing a product that is innovative is very different than developing a product that solves a defined purpose. With that perspective in mind, we ran a lot of ideation workshops. Prior to these workshops, we trained our customers and our internal participants on what generative AI can do today. Obviously, we did need to put some constraints in place to identify where the technology was too far off, and we needed to push those use cases to the future.
We then brought the teams together and explored what finance professionals spend time on and how we could make their days more efficient and more productive, while making it easier to do their jobs and obtain insights in a timely manner.
Leigh Pepper: It feels like with AI, you can use that analogy that says if you’d asked people how they wanted their mode of transport to evolve in the future, they’d have asked for a faster horse rather than imagining the idea of a car.
I think we’re at that point in product innovation, particularly around Gen AI, where product managers and product professionals are having to do all of those things that you’ve just outlined. They have to look at what the finance user is aiming to achieve as opposed to time and motion reviews of how to incrementally improve the current state. It feels like we’re at a stage where we’re going to see a huge leap forward in terms of the automation and optimization that Gen AI and Copilot can deliver.
Gurkan Salk: That’s a really good analogy, Leigh, and represents our thinking at Microsoft.
Mark Aubin: As you’ve been working with your charter customers and internal finance teams, have you seen any tangible benefits?
Gurkan Salk: The initial use cases were around data reconciliation, which is and has been traditionally a very manual and time consuming process. The feedback we received from our treasury team was that accounts receivable reconciliation capabilities have helped them eliminate the time it takes to compare data across sources, saving an average of 20 minutes per account. Based on pilot usage, that translates to an average of 22% cost savings in average handling time.
Mark Aubin: That’s an excellent result, and certainly supports the argument for how AI can increase user productivity. Can you give us some insight into what’s next for Copilot for Finance? A sneak peek into the roadmap perhaps?
Gurkan Salk: One of the next use cases is variance analysis. It’s a process that’s still very time consuming for finance professionals. Think about that process of matching your forecast against your actuals – how do you determine why you missed your forecasts? It requires finance teams to look at market conditions. Is it the recessionary period that impacted the forecast? Is it our sales execution? Is it a product issue? Teams have to go into several financial systems to understand the cause. So, we’re leveraging advanced data analysis feature by Open AI and that data analysis engine to turn that process from hours or days into minutes. Teams can collect all that data from different systems, input that into Copilot for Finance and, in a matter of minutes, get the root cause of why the forecast was missed.
The way we’re identifying and prioritizing these use cases is that we are trying to meet the user in the flow of work. If the user spends most of their time in Excel, which we know finance professionals do, those are the processes that we’re looking to light up in Copilot for Finance.
Mark Aubin: How does the broader Microsoft ecosystem add value to Copilot? Or conversely, what value does Copilot provide the broader Microsoft ecosystem?
Gurkan Salk: If you think about it, it starts with the very bottom layer, with that large language model and working with partners that provide AI capabilities and lighting them up in Azure as an API so that our customers know that they can trust this API, and that Microsoft will never share customer data externally or add it to these large language models – they know that this is a secure platform that our Copilot offerings are built on.
We then have the platform capabilities like Azure AI Studio and Copilot Studio. Those are the platforms we have developed for our partners and customers to develop their custom GPTs and to build their own Copilot capabilities. Microsoft uses the very same platforms to build our first-party offerings like Copilot for Finance.
Leigh Pepper: At Aptitude, we’re excited about using some of that tooling. We can see how the data that we hold within Fynapse – very rich, very well-structured financial data – can be that jet fuel that enables enterprise accounting clients to use Copilot in their flow of work, as you say, rather than having to leave the application to navigate other systems.
Gurkan Salk: I think there are a lot of partnership opportunities as we think about these Copilot scenarios that leverage the analytical data that is held in the Fynapse accounting hub and subledger. In those cases, we can leverage Copilot capabilities in Fabric, Power BI and Excel.
Leigh Pepper: Absolutely and I think the partnership with Dynamics 365 and the ability to actually drill back from the data that’s in Dynamics 365 all the way back to the lineage and the source data that sits within the Aptitude Fynapse system provides the opportunity to leverage that richer data set.
Gurkan Salk: That’s exactly right. I think that fits into our vision of autonomous ERP very well.
Mark Aubin: Leigh, you spoke about the role of the Microsoft ecosystem and the exciting things it can unlock for partners like Aptitude. Can you go a little bit deeper on how Aptitude is aligning our strategy and roadmaps and how we are looking at using solutions such as Copilot for Fynapse across the Aptitude portfolio?
Leigh Pepper: I think one of the things that Gurkan said earlier is that it’s really important that we enable finance professionals to work ‘in their flow’ or, in other words, within the tooling that they’re used to working in, whether that be Dynamics 365, Excel, or something else. If we can manage that data effectively and get it into an interface that users are very familiar with, one that’s in the flow of their work, I think it will help to drive high levels of efficiency for our joint clients.
We’re prototyping use cases using some of the Copilot Studio tooling, which really helps in a number of ways. First, it helps to simplify the implementation of new accounting platforms and ERP systems because we know that this process can be very lengthy and very costly. Using AI to look at the accounting rules, calculations and flows that they have in their current system and suggesting what that should look like in the Fynapse platform is a way that we believe can significantly cut down implementation time for clients. As an example, we’ve been using some of the Microsoft tooling in a few proof-of-concept projects and we’ve seen the time to implement go from a typical three months down to three days.
The other area, as you mentioned, is around data analysis. We can process the data into the accounting entries that are required in a highly scalable, volumetric way, that is very, very efficient. Then, we can put it into a database that is able to handle lots of different views over that data, allowing a user to apply multi-GAAP analysis and view different balance types off of a single data set. Finally, we can use Power BI and Copilot capabilities to build out some of those AI scenarios. It’s leading to some really good proof of concept insights.
For example, we had a case where we built out some analysis using the last three months of financial data that had come through which showed a big anomaly in one of the offers and promotions. A simple query over gen AI, using Microsoft Copilot was able to identify a particular issue on a single transaction. Then, because of the interconnectedness of Fynapse and Dynamics 365, instead of that just being at the very aggregated level, we were able to drill all the way back to that source transaction within Fynapse for the user to look at. This saved someone from having to scroll through hundreds of thousands of entries to try and reconcile the error.
Mark Aubin: That’s the key. Being able to detect those variances and those anomalies almost in real time rather than at a specific period end. Bringing those AI capabilities more upstream to the point of contract, the point of transaction will make companies much more agile.
Leigh Pepper: I agree. Having the whole architecture AI-enabled, from the source systems through the accounting hub processing, into the data and analytics layer and then out into the generative AI models allows for that semi real-time capability with anomaly detection before period end close which is where it really adds value. The partnership and the solution we’ve built, and what youguys are building on the Microsoft side, can really transform how finance departments are thinking about their operating model and how they move to autonomous.
Gurkan Salk: It’s exactly what we want to see. Microsoft generative AI creates value for the customer not only by saving the user time, but also by allowing the organization to act quickly to address something that could cause them to lose money or miss an opportunity in the market.
Mark Aubin: What steps should an organization take when they’re thinking about deploying Copilot? What do you think they should focus on? There’s always that intersection of low hanging fruit and what’s achievable versus pursuing true transformation.
Gurkan Salk: We think we’re at another big paradigm shift where generative AI will transform the way software works. We’re getting away from those button clicks and forms over data, to a world where the user interacts with the software much more freely, much more efficiently to get the results. In the case of Copilot for Finance, we’re transforming the way finance professionals work.
If you think about Microsoft’s vision of autonomous ERP, what we have today are Copilots that assist users. But as we progress in that journey of autonomous ERP, we want to minimize the interaction points that the user has with the system. We want the system to help you so that they don’t have to do the mundane tasks, the system does that for you.
I think in the short term, it should be all about what new capabilities you are able to do in Excel that you couldn’t do in the past. But finance should also be able to see the vision of where we’re going with these generative AI capabilities in the future so that the users are prepared and understand the journey.
Mark Aubin: That’s why I think the genius of the Copilot name is so perfect because it truly is a copilot for the user. It’s empowering finance users by placing the information they need right in front of them.
Leigh Pepper: I agree, the name is perfect. At some point we’re going to see this inflection point where you’ll access your primary interface, for example Excel, and then you click Copilot to help you. Then I think one day we’ll suddenly find everyone is just going into Copilot and accessing data and solutions from there – it will be interesting to see how that adoption curve progresses.
Mark Aubin: Excellent. Thank you both for your time and I am very much looking forward to seeing where 2024 and beyond takes us.
Gurkan Salk: Looking forward to the continued partnership with Aptitude in the Finance space!
Previously published in CFO Futures: An Autonomous Finance Magazine (Spring 2024)
