ork.AI
This blog post is Part 3 in our three-part series that defines the AI Agents by WorkFusion. In Part 1, we provided the highest-level definition: Digital co-workers that decide, act, and communicate.
In Part 2, we explained that WorkFusion AI Agents are valuable right out of the box because of their three key attributes: Pre-built, Explainable, Controlled.
Here in Part 3, we explain AI Agents as seen through the eyes of financial crime compliance operations leaders.
They are workforce augmentation
Each WorkFusion AI Agent is a focused, singular financial crime compliance process—automated and personified. Whether for KYC (know your customer), transaction screening alert review or transaction monitoring investigations for sanctions or AML violations, adverse media monitoring, or other core FinCrime processes, they are fully digital members of the team who work alongside real-world colleagues to:
- Reduce manual effort and workload
- Find the ‘true positive’ needles in the haystack of false positives
- Efficiently expand the overall capacity of any compliance team
- Deliver significant ROI
They supercharge people as well by relieving them from mind-numbing and repetitive activities such as data collection, document handling, and false-positive clearing. It almost goes without saying that supercharged people perform better when taking on more strategic and fulfilling projects.
For their own part, AI Agents never tire, enabling them to drive results faster and continuously. They deliver measurable improvements in team performance across financial institutions of all sizes.
Stewards of valuable data
In the world of FinCrime compliance, two major issues surround the data used by operations teams. It must be good and workable data, and the data’s security must be ensured while in use.
WorkFusion AI Agents ensure good and workable data
To be effective, AI thrives off data, and bad data can lead to AI making the wrong decisions. For over a decade, WorkFusion has been working with massive volumes of data—especially data related to anti-money laundering (AML) and other areas of financial crime prevention. So, our AI Agents incorporate pre-trained models for multiple use cases to make immediate impact, while their machine learning features continuously learn more about a specific institution’s data.
As for the workable nature of the data in use at banks and other financial institutions, AI Agents help them maximize use of the data to which they have already purchased access. Often, organizations do not maximize the value of the data they have purchased. By aggregating data from assigned sources and folding it into their data capture and analysis steps within a process, the AI Agents make the data more workable for compliance purposes.
They keep data secure and useful
In terms of data security, due to the popularity of LLMs like ChatGPT, many people assume that LLMs are always accessed via the cloud and that data used by AI must move outside of the organization’s firewall, becoming insecure. However, cloud-accessed LLMs are not the only AI option. LLMs like Llama (by Meta) and Mistral AI can be installed on-premises and run within an organization’s firewall. In fact, multiple customers of WorkFusion run our AI Agents inside their firewall to perform their roles.
While we will not risk revealing how our security operates and what specific products we use, WorkFusion has (and follows) a best-practices security posture and technology to support it within cloud, hybrid cloud, and on-premises deployments. In addition, WorkFusion never takes possession of an organization’s data—using it only within a process and never keeping it. When AI data remains secure, FinCrime compliance teams gain the benefits of AI without the risk of exposing critical data.
Ultimate data usage from modern and legacy systems
Customers of WorkFusion control which data sources an AI Agent may access. As such, an AI Agent cannot ‘go rogue’ and pull in new data on its own. But the main benefit that banks see when using an AI Agent is that they gain from all the systems they have implemented over the years, much of it not designed with AI in mind. Yet now they can let those systems’ data remain in place while taking advantage of it via AI. Typically, application programming interfaces (APIs) make it simple to integrate AI with older systems, moving data between existing systems and AI.
Customer data resides within many systems, often ten or more. To ensure you have comprehensive compliance data, your AI Agents should have connections to external data sources, news sites, and newer, unforeseen sources. By no means an exhaustive list, following are examples of systems and information sources which WorkFusion AI Agents connect with to optimize compliance processes:
- LexisNexis, Thomson Reuters, Moody’s and additional sources for enriching data
- NICE Actimize, Verafin, SAS, Oracle SCM, Fiserv, FIS and other transaction monitoring software
- Case management software
Integrations beyond APIs
Compliance teams recognize that data lives in many places, and there is little understanding regarding the varied access methods (beyond APIs) needed to tap into it. Certainly, APIs are a piece of the puzzle, but agents need to access data within documents, UIs, emails, and other places. So, our AI Agents have been designed with features and capabilities for employing multiple methods to access a variety of data sources.
Mature packaged processes, not custom builds
Our AI Agents are smart, arrive pre-trained, and collaborate with your team when they determine that a decision requires a person’s nuanced understanding of the issue at hand—not to be made automatically. In this way, FinCrime compliance professionals need not delve into technology, nor act as data scientists or machine learning engineers. AI Agents are designed to allow FinCrime specialists to focus where their crime investigation skills are best utilized—evaluating complex cases.
You can even be a technology laggard. This is why large banks and smaller FIs can both leverage AI today. It used to be that smaller banks lacked the staff to work with AI. But with AI becoming simplified via AI Agents, WorkFusion counts among our customer base a blend of both large and mid-sized institutions.
Explainable AI that supports MRM
Each AI Agent incorporates AI and ML in a “glass box” model which any financial institution’s compliance team can use, understand and explain. This supports model risk management (MRM) and leads to rapid acceptance of new AI-based solutions by senior management and model trust by the regulators who review an institution’s model risk.
Here’s an example: Multiple banks use Tara—our AI Agent that automates transaction screening alert adjudication for sanctions. Tara facilitates 70%+ reduction in the manual disposition of false positive hits on millions of transaction alerts each year. To make ‘her’ automated decisions, Tara compares the data she accesses against her decision matrix to determine whether an alert is a false positive or not. This decision-making is driven by a machine learning ensemble model and a supporting rules engine. The ensemble model is actually four AI/ML models. Per the FDIC’s supervisory guidance, the four AI/ML models employed by Tara incorporate the three core model components of Inputs, Processing and Reporting—with full explanations, audit trails, and reporting that address each component.
To further optimize AI explainability and minimize model risk, Tara (like all our AI Agents) is put through intensive and continuous testing of the models that underpin all decisions. This multi-stage model testing methodology ensures that final decisions (aka “outcomes”) are accurate and robust.
We hope you have enjoyed reading this three-part series and have developed a solid understanding of what WorkFusion AI Agents are and how they dramatically improve FinCrime compliance operations. To learn more, request a demonstration today.