California-based Talisai answers this question with its model and data governance tool that simplifies the adoption and operations of AI and advanced analytics in enterprise risk management and compliance domain within regulated and highrisk industries, especially regarding the supply chain risk, people risk and cybersecurity. And with the help of the company’s heuristic non-AI algorithms and model parsing capabilities, clients gain the evidences and data traceability behind decisions, which can be used to compare the decision-making of the AI algorithm and understand where it stands with respect to intelligence training. “Our tool allows business users to understand whether their AI is trained enough to be used in real-world scenarios, which drives better transparency, traceability, auditing, and protection,” says Joonho Lee, co-founder, and CEO at Talisai.
Further discussing the competencies of the algorithm governance tool, Lee elaborates on the solution’s supervisory and preferred modes. In the supervisory mode, Talisai’s tool considers the decisions made by the AI-based algorithm and compares it with the heuristic non-AI algorithm with supporting explanations. This parallel association between the analytics and the explanations allows Talisai’s clients to determine training feedback more accurately and understand where they stand in their AI training journey. Subsequently, when the AI model gets smarter and starts making decisions close to the business analyst’s qualitative assessment based on the ongoing results and explanations, the tool goes into its preferred mode, where it considers AI-powered decisions to be right always unless its results go outside of the determined risk boundary. In the preferred mode, the Talisai’s tool is also on the lookout for significant deviations from the decisions made by its heuristic non-AI algorithm running in parallel. As a result, in cases where there are potential anomalies or bias, the results are prevented from flowing upward to any business decision-making process until investigated.
This enables realtime risk prevention for companies. Developers can analyze the reasoning of the AI module and gain a deeper understanding of how to train the algorithm across its maturity phase. Concurrently, Talisai’s tool boasts a vertical lineage capability, which creates traceability from the bottom line datasets that reach the analytics module to the key business metrics that drives the final decision in a way that is 100 percent understandable to business leaders and C-suites.
Our tool allows business users to understand whether their AI is trained enough to be used in real-world scenarios, which drives better transparency, traceability, auditing, and protection
With such unmatched capabilities, Talisai has ignited several success stories since its inception in 2018. In one instance, the company assisted a client to improve their security and privacy programs to be more preventive and proactive than the current detection-based system could have. Similarly, Talisai has helped its clients in enhancing supply chain risk management, and complying with the ever-evolving regulatory mandates with respect to their data privacy, security, and more. At the same time, the company also deployed an intelligent data analysis feed providing risk severity of the pandemic, weather, natural disaster, and social unrest, all of which could help clients proactively mitigate the potential impact.
Looking ahead, Talisai plans to enter the crypto exchange and DEFI arena by leveraging its team’s extensive financial services expertise and developing financial regulatory compliance modules such as Anti-Money Laundering (AML) Know-Your-Customer (KYC) by leveraging the same core capability, Accountable AI. With this tool, users will gain the ability to have an ongoing evidence chain that will allow them to quickly detect potential suspicious activity or early warning signs of discrepancies in their chain of operations. From an expansion standpoint, the company aims to security, risk, and compliance sectors across all industries. “We will continue helping our clients in mitigating their data issues, managing their AI governance structures, and enhance their understanding of AI-based decisionmaking to successfully adopt and gain the desired ROI from their AI investments,” concludes Lee.