The Center for Data Innovation spoke with Eyal Feldman, co-founder and CEO of Stampli, a California-based fintech company that uses AI to streamline accounts payable processes for a range of enterprises. Feldman spoke about how an accounting error inspired him to start Stampli and how Stampli’s AI assistant, Billy, helps companies make their accounting departments more productive.
Martin Makaryan: What inspired you to start Stampli?
Eyal Feldman: Starting a company was not an easy decision. After gaining experience with enterprise resource planning (ERP) and document management systems, I thought there must be a more efficient way to connect people, processes, and documents. I wanted to use technology to break through ceilings that often limit process improvements, especially in large businesses. After exploring various options, I chose accounts payable (AP) processes because this is an important function for any enterprise—ensuring that accounts are in order and that you are processing invoices correctly, for example—but an area that needed more innovation.
The push for me personally came when I almost lost business in my previous workplace due to an invoice mistake. The mistake was fully avoidable and I realized the complexity of invoice processing. An invoice can go through a number of different channels depending on the type of organization, and this makes effective collaboration across departments very important. It was an epiphany moment that showed me we could make a significant difference.
Makaryan: How does Stampli automate AP?
Feldman: AP should be a collaborative process, and Stampli provides organizations with a software that allows them to turn each invoice into a landing page where all relevant parties can work together. On this landing page, we provide all the necessary information: discussions, questions, collected data, relevant documents, etc. This central hub allows customers to reduce the time accountants spend on invoice processing, reduce bottlenecks, and avoid human errors. We also ensure seamless integration with the customer’s accounting environment, either in the cloud or using their on-the-premises hardware. We offer native integrations for more than 70 ERP systems, including complex ones. Our goal is to avoid disruption and ensure that Stampli feels like a natural extension of the organization’s existing system.
We have also incorporated AI to predict next steps and save time for our customers. We have created a generative AI assistant, Billy, that automates many time-consuming accounting tasks, like summarizing text and answering questions related to an invoice. Billy frees time for users to focus on more important work and avoid simple human errors in accounting. Billy can also suggest cost allocations and predict workflow steps to ensure that an invoice goes through all the stages based on an enterprise’s AP policies. Another key feature that makes Billy very useful is that it learns from mistakes and corrections accounting staff may make.
Makaryan: What success stories show how Stampli’s automation has led to significant improvement?
Feldman: We have had numerous success stories. For example, one of our customers reported that their invoice processing time decreased from 40 hours per week to just a couple of hours, with accuracy approaching 100 percent. Many customers have told us they were able to scale their business without adding more staff, using their existing team for more strategic tasks. Perhaps the most exciting feedback I am hearing is that enterprises that use AI tend to have more appreciation for their accounting and finance departments. The improved collaboration has enterprise-wide benefits, as AP processes involve many people outside of accounting departments.
Makaryan: How do you use data to improve your product?
Feldman: We use data in two main ways: First, we have a sophisticated approach to creating, distributing, and verifying our own data and insights. We have a team working closely with me to ensure that different departments have the right insights to focus on our priorities and use customer feedback to tweak our offerings. Second, for improving our AI systems, we leverage the full workflow data from our customers. Understanding each step of the AP workflow for our users allows us to effectively reverse engineer what a user would do in a given situation, in turn helping us create AI systems that will perform tasks that our customers are most likely to need help with. Using this comprehensive data is why we are well-positioned to develop sophisticated solutions in the accounting space.
Makaryan: What is a challenge you are facing as a data innovator?
Feldman: A recent challenge for me has been the market noise around AI. The rise of generative AI has been very exciting and promising for many fields, but many companies claim to be “leveraging AI” when they are only scratching the surface of what is possible with this technology. Our challenge is to educate both our existing and potential customers and stay focused on bringing real value, not just what looks good in marketing slides. We approach AI like hiring an employee—it is not about making big claims but about demonstrating real value to customers.
