Learn about Avani Desai, CEO of Schellman, an IT compliance and cybersecurity audit firm. One of our 50 Women in AI. We spoke about how she confronts the challenges she sees in auditing and in the field of AI. During our talk, Avani identified two main challenges: Keeping our data safe and finding the best people to do so.
Energetic and enthusiastic, Avani spoke openly about the challenges that auditing faces. Rather than go with the status quo, she and Schellman are shaking up the auditing industry, a sector that traditionally required hours on end of manual labor. Rather than go with the status quo, she and Schellman are shaking up the auditing industry, a sector that traditionally required hours on end of manual labor, scrutinizing data and looking for errors. And even with high focus over long hours that tossed work-life balance out the window, the risk of human error loomed high. But now, AI powered audits allow Avani and her team to deliver better quality and more effective, comprehensive audits.
AI technology relieves Schellman employees from tedious hours of administrative tasks that bogged them down, catches more errors and frees auditors up to spend time in higher value areas.
How to keep up with monitoring absurd amounts of data?
2.5 quintillion bytes of data being created every day, 90% of the world’s data has been created in the last two years alone. A staggering figure, it is expected that the volume of data is to double every two years. ( 1 quintillion = 1 billion gigabytes)
Avani noted that business’s increasing use of new technologies puts even more demand on those who have to audit the data, to keep our data safe across countless new platforms and tools:
With the emerging technologies today, we have blockchain, we have machine learning products, we have AI products. All of those need these audits, and we just don’t have enough people to keep up with it.
Background in IT Risk Serves as Foundation for Data Driven Auditing
Avani came from a big four accounting background, working in IT risk. As Schellman’s CEO, she’s helping their clients keep up with the growing demands to keep our data safe. Their work puts their client’s data and cyber security to the test. Their audits work to answer such questions as:
- Does their data provide the info it needs to? Is there bias in it?
- How is it collected and stored? Do they follow legal and compliance rules?
- How do they keep the data safe from hackers?
- Has the data been manipulated for purposed of fraud?
Keeping her client’s data safe, (which also means keeping our data safe, since her clients span across industries we all interact with, from healthcare to finance). The old school way of auditing involved selecting a small sample and verifying it. But being only a sample, there was a chance something important was missed. And now there is a lot more data to review, how can this be managed effectively?
AI powered audits now provide a way forward.
Evolution from manual reviews of sample data to continuous automation auditing of all data
Along with taking a lot of time to review just a portion of what needs to be examined, manual auditing is subject to human error. Typical sample sizes contained just 10- 15% of all data that the company managed. What could auditors be missing in the remaining 85- 90%? How could they ensure that ALL of the data is safe? Avani commented on this challenge:
So if you think about the audit profession in general, it has been a very manual process and has been for hundreds of years. You’ve probably heard people say, oh, I have to look through thousands of pages. I have to tick and tie, I have to get the ledger and make sure that it adds up appropriately. So what’s happening though, in a world where technology companies or any company is using an ERP (enterprise resource planning) system, is using cloud products, what do you do when there’s increased automation? You have to make sure that your audit evolves in the same way. So a great example is if I am typically selecting a sample during an audit, I can’t select everything, right? There are a hundred thousand items that are out there.
I’m going to take a sample of something and I’m going to review it. But there’s so much data that is out there. How do you get completeness and accuracy if you’re only looking at a portion of it? Do shareholders feel confident if you’re only looking at 10% of the population or 15% of the (data)?
AI aids auditing clients increased trust for end clients
With the help of AI automation, rather than selecting a sample that captures only a small percentage of the total data, all data can be reviewed continuously. This reduces the risk of missing something important resulting in increased trust among clients and shareholders:
So now we can actually look at machine learning and AI enabled audit tools. We take all this enormous volume of data to be analyzed because what we want to find is that anomaly or some insights, some patterns. Then our auditor can take a look at the output to really determine if that’s going to be an exception, a true anomaly.
While AI auditing has generated enormous strides in effectiveness, people are still needed to work with the AI and dig deeper when problems are signaled. And this brings us to the second big challenge Avani faces: finding the talent needed to work with the technology.
Competition for top talent to attract the best brains to auditing
Ask a kid what they want to be when the grow up: Firefighters, dancers, astronauts, doctors, or pro athletes in their sport of choice, something deemed both exciting and honorable. (Accountants and auditors usually don’t make this list) But many of us outgrow our childhood dream professions, go to school, and consider how we can best earn a living.
Some bright individuals choose the reliable, if not overly exciting profession, of auditing or accounting. But smart graduates are choosing this profession less and less, lured by more intriguing, innovative work in tech and startups. They seek to put their skill sets to work to create something new and grow, rather than overwhelm themselves with time intensive, boring administrative tasks. Auditing’s reliable if unimaginative image, is not doing the field any favors.
The reason most people don’t go into auditing and accounting is because the perception is 60, 70 hours work weeks. You’re sitting in a room with 20 other people, you order in lunch and you just go through documentation all day.
The audit profession is seeing less and less people interested in auditing. They say about a million less individuals are going to school to be into accounting. And the cost, or really the war on talent is so prevalent in our industry. And so most audit firms, especially the top 100 audit firms, are focusing on how do we make the audit engagement more efficient?
It’s clearly that the image has been tarnished over the years. Usually people who would graduate with an accounting degree, 40% would not get their CPA, and now 60% aren’t getting their CPA. We’re starting to see the trend. People are going towards more tech enabled organizations where there’s not so much red tape, so much manual work. So I think we need to really change the narrative of how can AI improve the cybersecurity audit space, how can AI improve the financial audit space, and how can you learn to be a better auditor but still have that work life balance?
AI to solve the problems of keeping data safe and attracting top talent
With two big obstacles: protecting our data and finding talent, AI serves as the proverbial stone for the these two demanding birds.
Unless you’ve been living under a rock, you know that AI’s applications already provide value across numerous fields, especially in releasing humans from performing repetitive tasks. And with much of auditing’s heavily manual processes, there’s ripe opportunity for AI innovation.
Auditing has never really embraced innovation, and this is finally embracing innovation. Auditors and accountants have been known for high quality. And at the same time, burnout in this profession is real, and I think this could combat that. (With AI) they stop focusing on tasks that are repeatable and burdensome, because the burnout really happens when you’re looking at the same thing over and over again.
Instead of looking at just fraction of the data that manual audits could cover, AI tools look at all of the data. These tools raise flags on detected anomalies, signaling auditors where to dig deeper. The tools free up auditor’s time to focus on higher risk (and more interesting) areas. AI tech doesn’t replace employees, but allow employees to accomplish more. AI tools also function as a foundation for more innovation, allowing auditors to develop scripts to improve the process.
AI automation, once set up allows clients to monitor activity in real time, quickly identifying issues and protecting the data from those who could use it for harmful purposes. Schellman is also building their own AI powered proprietary tool, Audit Source, which reduces the amount of time spent manually gathering, tracking, and reporting upon audit evidence requests.
The end result is a better tool and more engaged employees and people starting to consider this type of work again.
Innovation through diverse backgrounds
Not just reserving innovation for those already inside Schellman, Avani is innovating in how she builds the Schellman team. Venturing beyond CPAs, she started including those with non accounting backgrounds to bring in new perspectives. She’s found employees from non-auditing backgrounds contribute new insights and improvements to the process.
We also look for employees in the nontraditional roles and teach them. So if we can, we go out to market with traditional roles like accounting, computer science. But what if we can find roles that are nontraditional, where people would be interested and open to learning? I think you’ll be able to get a bigger pool from there. The other thing I think is typically when we go out to recruit, we’re looking for people two to five years outside of college. But maybe we start looking at work experience of ten plus years and veterans and so forth.
She talks about Schellman’s X Program, which seeks out recent college graduates, but with different backgrounds than they previously targeted:
They’re not all computer science majors, or science majors or management information systems majors. And what we’re starting to see is that we have this really diverse and inclusive cohort that is starting. They’re fueling a lot of productivity and innovation and we’re seeing better service.
Helping techies non-techs alike, acquire the needed skills, Avani found success by breaking down complex material into simpler, digestible chunks. And this same approach can help increase the lack of diversity currently in the AI field.
Overcoming the technical barriers to entry in AI – Bringing more women into the male dominated fields
Avani works to increase diversity in these careers areas that still see a large gender imbalance.
There aren’t a lot of women in AI, there’s not a lot of women in technology, and there’s not a lot of women in cybersecurity. And that is one passion of mine – how do we change the role of women in AI and cybersecurity and so forth? And I think it’s exposure at a very young age, women have a lot of great talents like multitasking and solving problems and looking at building algorithms. Women are great at those things. It’s part of our ethos. I’ve talked to a lot of women who say, well, I don’t want a code and I’m not a data scientist. And I said, but that’s not what AI is. You don’t have to code and you don’t have to be a data scientist.
To bring more women in the field she further stresses the importance of starting young. Avani works to attract young girls and show them the potential of AI. She cites gaming as one way to demonstrate the AI career path:
And starting early, you have to start early, like elementary school and telling them what AI is and how their everyday lives are affected, and about gaming. I’m a big believer in gaming. It helps you be a better developer, a better cyber security specialist, and so forth. I let my kids game often. Gender diversity is important in creating the most effective algorithms
The other thing I think is gender diversity really matters, especially in this unconscious bias or bias that we’re starting to see. I mean, that is another con with AI. Depending on how you build your algorithm, are you going to have bias associated with it? And I think you need to have diversity to make sure that the benefit is there for the organization.
AI proves successful to protect our data and attract talent
Avani’s approach to AI and diversity seems to be paying off. She and Schellman have developed a more efficient and effective process to keep our data safe. She’s helping her field break out of a rigid and outdated view of that was damaging the profession and not delivering the highest value to clients. She’s accomplishing this thanks to AI and building a team that’s diverse in gender, experience and education .
Schellman has received numerous awards across different categories, from top places to work, to fastest growing and use of technology. It seems with AI, Avani has found the sweet spot to confront her two big challenges resulting in benefiting her employees and clients.
You can contact Avani at: firstname.lastname@example.org, learn more about Schellman, and follow here on Twitter. and follow her on Twitter: @AvaniDe