Armanino’s Ryan Prindiville recently connected with Dr. Alok Aggarwal, CEO of Scry Analytics, to discuss how their team is creating innovative solutions to augment automated real-time data analysis with domain expertise.
Ryan: Tell us how Scry Analytics was created.
Alok: After working in machine learning and artificial intelligence for much of my career and writing computer algorithms to increase manufacturing productivity, I decided to start a company with the focus on the idea that AI will be ubiquitous in everyday life within 20 to 30 years.
Just as machines surround us in buildings, factories, and vehicles, AI applications will be fundamentally embedded in daily life. We intend to be part of that by building apps that make up the underlying functions of global companies, just like Intel chips. Scry Inside, if you will.
I believe that in the future, the special-purpose apps like Scry’s quality assurance app for IT and software development will be invisibly powering businesses.
Ryan: What trends have emerged in the past two years that you think will affect your work in the next two years?
Alok: I have seen people express concern that increased AI will cost jobs. I believe that embedded AI has little to do with performing functions that humans do. For instance, sustainable energy companies will offer more jobs in the future, but they will need AI’s help in maximizing power storage gathered from micro grids. AI algorithms can offer accurate usage predictions to ensure the most efficient distribution of stored energy. Imagine using weather forecasts to distribute energy where it’s most needed or using COVID data to know that people in a certain town are working from home.
AI will also be instrumental in smart cities in that an algorithm can calculate thousands of variables to predict, such as how long it will take people to travel from one point to another. Supply chain management will also be a huge user of AI to calculate time and distance between commodity deliveries, and that information could even affect investors hedging on companies.
Ryan: Can you talk a bit about how AI has changed Wall Street banks such as Goldman Sachs and has moved their business models to be less about financial management and more about data science? They now have more data scientists and software engineers on staff than they do finance, accounting and wealth advisors.
Alok: That’s exactly right, and the same, I think will happen with consulting firms such as McKinsey. For instance, they have so much data about a company and all its competitors that it blows up consulting issues and becomes much more about the data itself. But whenever things get blown up, new things emerge, right?
Fundamentally things will change, be it doctors or consultants. And people who adapt to the change will do well. It’s not that jobs will be lost, but that so many jobs will be created, especially when AI gets embedded in climate change solutions and we figure out all the things we can do to reduce our global carbon footprints.
Ryan: How has the AI strategy for businesses changed?
Alok: I compare it to the industrial revolution that changed which countries would go up and which would go down. I think the same will happen with AI. It’s the same kind of revolution and some countries at the top now will not remain so in 30 years. It has to do with how countries view data privacy, how much data they have and how they can use it. Countries like China and Singapore can use the data more freely, or India, Saudi Arabia and the United Arab Emirates. It’s going to be an interesting 30 years.
Ryan: Yes, that’s interesting: lack of data privacy restrictions is actually increasing the size of the data set available to drive analytic capability. I think that qualifies itself as a megatrend.
Alok: Absolutely. And people don’t realize it. I tell people, “Look, I did my PhD in machine learning algorithms. I’m an algorithms guy.” And I would also say that more data generally beats algorithms because you can train more with more data.
Ryan: Tell us what some of your apps are doing for companies today.
Alok: Collatio is all about the digitization and reconciliation. This app takes data that is in PDF machine-readable format, a scanned format, audio data — and converts it. If the data is in PDF machine-readable format, we don’t have to send it through any optical character recognition. It gets better accuracy by doing other things, so it converts it into an electronic format. After the conversion, it learns from that document type.
For instance, if it’s a W-2 form or a balance sheet, it will begin to figure out the possible formulas within the document, which columns and rows are equal or have relationships to each other. From balance sheet data, it can help determine a company’s financial health to determine loan eligibility.
Ryan: So it can more quickly absorb the P&L and any other required documents, and compare it to the larger populace to determine the likelihood of good performance?
Alok: Exactly. Similar to a mortgage or auto loan approval that can take several weeks, it could narrow that timeframe down to a couple of days.
I think the trend going forward will be AI that can read tax returns and W-2 forms and have loan paperwork done in a fraction of the time it takes now. This type of efficiency can potentially affect companies, cities and perhaps even countries. So Collatio does that.
Another app, Anomalia, is all about figuring out fraud detection in ACH transfers, debit transfers, mobile check deposits, wire transfers and money laundering. Conflict of interest is a very big area for banks, auditors and law firms, to name a few. Figuring out conflict of interest is difficult because you not only have to check a particular entity, you also have to check associated entities, parent companies and such. It’s a complicated business.
Then, figuring out fraud — if an employee has gone rogue, like the Wells Fargo fiasco where branch employees were setting up bogus accounts. Anomalia helps with that.
Ryan: Scry works with a lot of banks and credit unions. What are some of the challenges of the smaller banks who don’t have the kind of profit margins as a Goldman or Citi? Do you think they will lag behind because they have the exact same challenges but not the same capital to do a proof of concept?
Alok: These smaller banks tend to be very fast movers, since they do not have as much administrative politics, nor do they have “not invented here” syndrome. And while they may not have the financial wherewithal to perform a proof of concept, if they have a product that is up, running and performing well, they can avoid a proof of concept altogether.
Anyone offering this type of product needs to be prepared to prove their accuracy right away. And between the makers and checkers — makers being the people converting the data — replacing the makers with software that eases the burden of the checkers.
Ryan: What advice would you give to the leadership of one of those small companies about making the move to incorporate AI into their business processes?
Alok: My advice would be to not even think of AI. You have a problem to solve, so what does it matter who solves it for you? You just want the solution. And if it’s a better, cheaper, easier and faster from a use perspective, then that’s the solution you should use. It doesn’t matter if it’s AI, MI, ML or any other abbreviation. Keep an open mind, and we can demonstrate how it works.
Ryan: So, don’t think about AI, do a proof of value on your own data and see the potential benefits. And maybe the expected result might be surprising.
Alok: Right, you have a business to run, so run it in the most optimized way you can. My second piece of advice is that if the problem is pressing and nobody has solved it, take the solution with a grain of salt and go for a proof of concept. But most small and medium-sized banks aren’t running into new problems that need solutions. Rather, they want cheaper, better, easier and faster solutions for already known problems. So proof of value is what should matter to most CEOs and COOs.
Ryan: What would you say to someone on the fence about whether to partner with an AI company?
Alok: When it comes to AI, instead of moving quickly, many companies in the U.S. feel that they can do it themselves. Of course, it can take them two years to do it and things may have changed drastically by then.
Ryan: Are there industries that you feel you might be a threat to now or in the future as you evolve?
Alok: There are a lot of people doing reconciliation and auditing. And I would argue that a lot of this work in the next five to 10 years will be done by machines. But if a company starts using AI for reconciliation, they can be ahead of the game and continue to employ those humans to tackle the next set of industry challenges. It could, in reality, make everyone more productive, not less.
Ryan: Our CEO has said publicly that we’re going to have to start rethinking the entire industry. It sounds like you’re on that same page.
Alok: Many, many services firms need to start rethinking how they do business, but it’s not a question of if, it’s a question of when and how fast. There are large consulting firms that have made a very good business by hiring the cream of the crop, the smartest people from Harvard or Stanford, etc. They learn the tricks of the trade and have residual knowledge from one company that will be used for the next company in that same industry, and so on. In future, algorithms will behave in a manner similar to such people and learn from new data and residual knowledge.
But the underlying thing is the data that is associated with it. That data will be sitting in these machines that will be able to figure out patterns much faster and better than any human.
Learn more about digital transformation and AI with Armanino.
