Nobody really likes being audited, but it surely is a crucial function to be certain that an organization and/or individual's practices and policies are correct. And like all jobs, someone has to do it.
Auditors typically spend their time examining an organization's risks and the way they control them – aka Risks and Controls. But writing down all risks and controls in a regular risk control matrix or Risk assessment matrix can take plenty of time, regardless that they are sometimes similar across different departments and firms, leading to a tedious, repetitive task – the type that seems perfect for an AI solution.
That's exactly the concept Audit Board, a 10-year-old cloud software company developing a web-based platform specifically for accountants. This week, The company has officially launched its advanced artificial intelligence capabilities Known as AuditBoard AI, it goals to dramatically improve the efficiency of audit and risk management teams.
“The biggest problem we see with our customer base is the increasing gap between the work that should be done and the variety of employees and resources that their audit team or compliance team has,” said Anton Dam, vp of engineering for data , AI/ML at AuditBoard, in a video call interview with VentureBeat conducted earlier this week. “For us, AI is what can step in and construct a bridge.”
A story of innovation in a staid industry
Audit board was Founded in 2014 by Daniel Kim and Jay Lee, each childhood friends and former auditors at PwC and EY. The current CEO is Scott Arnold.
The company has experienced rapid growth and adoption among the many audit, risk, sustainability and compliance teams of enormous corporations and now boasts serving greater than 2,000 clients, including nearly 50% of the Fortune 500.
An enormous a part of this was the cloud-based web app AuditBoard risk management platform, formerly referred to as SOXHUB, which streamlines the creation of risk assessments, audits and risk assessment matrices.
Before AuditBoard, many auditors relied on off-the-shelf spreadsheet software that that they had to manually customize to their needs, often wasting invaluable time simply setting it as much as fit their workflows.
Instead, AuditBoard was introduced to offer them with a customizable dashboard and creation tool tailored to them and their commonest functions, including issuing document requests to other parts of the audit firm, sorting audits by different categories, and viewing Real-time status updates on exam progress and allows for @mention of comments and threads, much like other cloud-based shared documents.
Now the corporate is moving to construct in much more automation because of the introduction of generative AI and huge language models (LLMs).
Speeding up exams with human-verified LLM results
Instead of an auditor having to manually enter a risk and controls description into their matrix in AuditBoard, auditors can click a brand new Generate button that uses an LLM under the hood to output really helpful language for the risks and controls.
The LLM examines the context of that specific item, including the associated documentation provided by the auditor or individuals audited, and produces a really helpful description. All the reviewer has to do is accept the result or quickly edit it and move on to the following description.
“We’re not only popping a chat bot modal,” Dam told VentureBeat. “This is built into the UX of our platform”
An AuditBoard customer told VentureBeat that it reduces the time it takes to jot down these repetitive description entries from 10 minutes to 1 minute.
“AuditBoard AI has been a game-changer for me and my team,” said one other comfortable customer, Melissa Pici, Senior IT Audit Manager at Syniverse, in a press release. “I can do things faster, which is a big capability multiplier. What used to take twenty minutes now takes just five.”
What's under the hood?
AuditBoard declined to specify which LLMs it was using for its recent AuditBoard AI functionality, but Dam said that Microsoft is amongst its AI partners and that the corporate shouldn’t be a “single-model shop” and is over-trained and refined coordinated models available. wherever that is smart.”
Additionally, Dam told VentureBeat that with the brand new AuditBoard AI, different models can be incorporated into a selected auditor's workflow depending on the function they desired to perform.
“We can be refining a model that enables us to adequately compare a framework requirement with a control,” Dam said, for example of what the corporate has been doing behind the scenes with LLMs.
“The key for us is: Does it work for the domain and is it top quality enough?” he explained.
The integrity and security of customer data is “inviolable”
Additionally, Dam identified that its clients had high data security requirements as a consequence of their operations in highly regulated industries, and explained that AuditBoard AI separated all client data and never merged it, meaning each client was free to make use of the AI even without using data They fear that something will leak outside of their AuditBoard account.
Customer data entered into AuditBoard is shipped via encrypted traffic to secure Microsoft Azure cloud servers for LLM processing after which back to that specific AuditBoard account.
“For us, the client boundary is inviolable, we don’t violate it,” explained Dam.
By automating routine tasks and providing intelligent suggestions, AuditBoard AI enables practitioners to create detailed risk and control descriptions, discover overlaps in work, and map controls across different frameworks. The solution is now generally available to all AuditBoard users and customers.
