Armanino’s Ryan Prindiville recently connected with Alex Hagerup (CEO) and Zak Stratton (VP of Product) of Vic.ai to discuss how they are striving to automate the accounting industry.
Ryan: Tell us the story behind Vic.ai.
Alex: Vic.ai is essentially building an artificial intelligence platform that is automating accounting, which is quite ambitious. But one module or skill set at a time, AI eventually will be able to perform accounting services at the level of or surpassing humans, and the people will be there for exception review.
Ryan: Things are moving so fast in the AI space. What are some trends and use cases that are affecting your product strategy?
Zak: AI aside, accounting firms are trying desperately to standardize. They end up logging into multiple platforms and doing things to 28 different workflows across 52 different clients. We’re finding ways to normalize the workflows so that instead of 28, you have three.
Ryan: There are a lot of different players who say they have AI solutions. How do you think about AI internally?
Alex: AI is kind of like getting into a self-driving car. You want it to be the best, the one that doesn't crash. So, you want to buy the best AI because it's an investment, and the best AI learns over time from your business. If you invest in a platform with a lesser quality AI, you're going to be left in the dust, and then it's hard to change two years down the road.
We worked on the AI piece for more than two years before we built any other part of the application. We started with a data set of over 200 million accounting transactions. And we did all the data analysis and all the AI modeling until we had something that worked. Then we started building the user interface and the mobile apps and all that. Everything we do is built around the core of the AI.
Ryan: What are some things businesses should look for when evaluating AI vendors?
Alex: The more questions you ask, the better. The most important ones are do you have humans in the loop, and where did the data set come from. To have AI that performs at a human level, you need a very large data set and a team of AI scientists and engineers. If a company cannot prove they have the team and were able, in a logical way, to acquire a large data set, then likely there isn't good quality AI there.
Zak: When someone's demoing their software, if it takes 20 seconds or less to accomplish something, that's AI. If they tell you that they send the data off to their back end and in 24 hours information will be extracted, that is not AI.
Ryan: What are the biggest misconceptions about AI that you’ve seen with business leaders?
Alex: A lot of executives do not realize how tremendously AI is going to change certain parts of their business. The automation is real and it's going to impact staffing and pricing. I don’t think they generally understand the magnitude of what's going to happen over the next few years. AI companies are driving real ROI and high automation for customers, and it's just going to grow exponentially.
Ryan: Is that misconception just in accounting or does it apply to your peer companies who are targeting other industries?
Alex: That’s the case broadly because we are in the beginning phase of this. Probably for a few more years, AI will be good at very particular tasks, very isolated work for the most part. And you have to be in the trenches and see that particular piece of work getting automated to truly understand the impact.
We had some very significant breakthroughs in the evolution of the AI algorithms in 2020, with some of our clients going from a lot of augmentation but zero true automation with AI all the way to being 70% fully automated with AI, so that the human component has been 70% replaced. It’s just going to continue exponentially.
Ryan: What are the risks of trying to do something yourself versus leveraging the established best products in the market?
Alex: If you're a large company with tons of resources and you have a very specific problem that doesn't lend itself broadly to other companies, then that may be a use case you should potentially try to solve yourself. But if you're looking at broader problems that can be solved with AI, you should never try to build it internally. The external solutions you can get will always be better because the vendors have the advantage of catering to many clients and thereby seeing more data.
Zak: The volume of data is fundamental. If you're solving something in-house, you only see your data.
Ryan: What AI trends are you most excited about, in technology or in learnings that you're making and applying?
Alex: When we launched in the U.S. in late 2018, there was hardly one accounting firm deploying AI for anything. Now there are at least 100 of them. So, the adoption is really exciting. Then you have the piece that's almost exponential in the progress and outputs of the AI itself. We’ve gone from a high level of augmentation, but no true full automation, to more of a balance between augmentation and full automation. Give it a year or two, and it’s going to be primarily full automation and exception handling from humans.
Ryan: What can't AI do?
Alex: That's quite a lot today, so answering what AI can do is easier. I think that's going to change dramatically over the next five to 10 years, but right now AI can be great at very specific tasks that are somewhat repetitive in nature and require reasoning or decision-making.
Ryan: Should accounting firms feel threatened? Are you eventually going to know more about their clients than they do?
Alex: No, our strategy is to work with and for accounting firms. They leverage our AI platform to become better accountants because they can do less repetitive work and more advising. And our platform is on the path, as well, to help with that advisory piece. We're not competing with accounting firms, we're just helping them become better.
Ryan: Are there any other topics you think are important in terms of demystifying AI for business leaders?
Alex: Just get started because there is minimal risk and an exceptionally large upside. Having AI deployed and generating ROI in one area of your business is going to open your mind as to how it could potentially improve other parts of your business, as well.
Ryan: What's your advice to other startups in the AI space?
Alex: Whether you're starting in AI or moving into that space, it is really important to attract talent. It's all about the people.
Ryan: I think that's a good takeaway to end on. The reality is that even the best AI in the world relies on the best people.
Learn more about digital transformation and AI with Armanino.
