Table of Contents
Key takeaways
-
Full Automation: Sensa Agents automate tasks like case summaries, SARs, and web research, freeing up investigators for higher-level work.
-
Improved Efficiency: AI handles large data volumes, detects hidden risks, and reduces false positives, leading to faster, more accurate investigations.
-
Customizable & Transparent: Institutions can personalize agents to their needs, with full control and transparent decision-making.
-
Continuous Improvement: Agents learn and adapt over time, improving accuracy and staying ahead of new crime tactics.
-
Cost-Effective: SaaS deployment lowers costs and scales easily, providing a high ROI for financial institutions.
The first Sensa Agents can autonomously summarize investigations, draft SARs, and conduct web research
As a pioneer in AI and fintech, SymphonyAI is committed to leading the charge with innovative Software-as-a-Service (SaaS) solutions and capabilities. One such example is Sensa Agents that use advanced agentic AI to automate and streamline manual investigation processes. Sensa Agents empower organizations to achieve significant operational efficiency savings while also improving risk assessment effectiveness and consistency.
The challenge for financial institutions
Financial crime investigation has never been more difficult and complex. Huge economic and social harm arises from money laundering, fraud, human trafficking, sanctions, and terrorism.
Investigation teams struggle with changing regulations, the need for real-time updates, siloed data, a lack of time, a lack of resources, and many more issues besides. The truth is that as finance has become more global, criminals have exploited these issues within financial services to enhance their activities. With rising costs, supply chain vulnerabilities, and increasing use of online channels, more than 70% of executives expect financial crime to increase in 2025.
Despite these difficulties, the tide is turning
Predictive AI and generative AI are helping more banks than ever before to fight back against bad actors. Absa, a leading South African bank, saw a reduction in false positives of 77% while a European bank operating across multiple countries saw average potential effort savings of approximately 20% in Level 1 and Level 2 investigations while also achieving significant cost savings of €3.5m a year.
But there is still more that can be achieved
What if AI could autonomously complete complex tasks, freeing up investigators to focus on doing what they do best – assessing risk. What if these AI ‘agents’ could all interact with one another, further automating workflows and improving their knowledge with every investigation? What if these agents were trained on your institutions policies and procedures, personalized to your institution’s unique processes, configurable to your every need, fully auditable with exceptional model governance built-in, and complete human-in-the-loop oversight wherever you wanted it?
The good news is that these AI agents are available now.
What is an AI agent?
Agentic AI is the important next phase of AI evolution, comprising two types— copilots and autonomous AI agents.
- Copilots are AI models that respond to prompts or execute predefined tasks, assisting human financial crime investigators. An example would be Sensa Copilot.
- AI agents, also referred to as agentic process automation (APA), work dynamically within predefined work scope and requirements. These AI agents can make decisions, plan, and learn from previous experiences. They can even collaborate with one another to seamlessly complete complex workflows. An example would be Sensa Agents.
Unlike AI copilots that support human investigators and perform their risk assessments, AI Agents can fully automate pre-defined tasks throughout an investigation workflow to remove the burden of highly manual processes that are inefficient, inconsistent, and prone to human error. AI Agents can even work together within complex workflows to automate the analysis of vast amounts of data, uncover hidden risks, and enhance investigator decision-making. As with all AI used in financial crime prevention, human oversight is always present as a safeguard.
Benefits of using AI agents in financial crime prevention
There are many benefits of using autonomous AI agents in financial crime prevention. These include:
- Enhanced capability: AI handles large data volumes, detects hidden risk faster than humans, and automates highly manual tasks like adverse media or background checks.
- Transparency and trust: AI agents provide explainable processes, work within clear parameters, and build confidence with stakeholders and regulators. AI can be tweaked to further enhance performance.
- Improved productivity: Using AI agents to automate complex investigative tasks, especially AML and fraud detection, freeing up humans for strategic work and leading to faster processes.
- Self-learning systems: AI Agents adapt over time, get better at spotting anomalies, and keep up with new financial crime methods.
- Adaptability: Autonomous AI agents adjust to new criminal tactics as they happen with the ability for real-time strategy updates.
- Superior accuracy: AI agents help to reduce false positives. This means less manual work and more efficient transaction monitoring and screening.
- Increased intelligence: Agentic AI can augment human investigators with analytics, quickly providing in-depth insights into patterns and vulnerabilities.
- Cost-effectiveness: Using autonomous AI agents significantly lower operational costs, which leads to faster compliance and a great ROI. SaaS-based deployment enables scalable adoption without costly infrastructure investments.
- Proactive risk management: AI agents prevent threats before they happen, helping financial institutions to avoid fines and reputational damage.
Learn more about the benefits of using agentic AI in financial services.
How do Sensa Agents work?
SymphonyAI is launching Sensa Agents to help financial institutions start enjoying the many benefits that AI agents provide. Designed to work in combination together, the result is intelligent workflow automation that truly speeds up financial crime prevention, making for a safer, more reliable approach to preventing bad actors globally.
The first three currently available are the Summary Agent, Narrative Agent, and Web Research Agent, and they all form part of a connected ecosystem with many more Sensa agents becoming available soon.
Maintain complete control
Using agentic AI removes manual processes but still ensures that organizations keep full control. Each Sensa Agent is personalized – trained on your policies and procedures, updates are actioned instantly. In short, financial institutions can customize their agents to work exactly how they require and can connect them to any internal or external data source via webhooks and APIs.
With proven experience at some of the largest financial institutions in the world, Sensa Agents may work autonomously but they also provide companies with complete oversight of their actions:
- Strict governance – Sensa Agent parameters are completely controlled and can be configured to align with organizational preferences.
- Human-in-the-loop – Human oversight and approvals at any stage of Sensa agent workflows with agents learning from user feedback.
- Transparent decision-making – Agents provide detailed descriptions to justify their reasoning and decisions.
- Fully auditable – Every decision, every action, and every outcome. Everything is recorded and explainable.
How Sensa Agents are trained
Sensa Agents learn using foundation model learning, reinforcement, instruction tuning, and goal-oriented planning techniques
- Industry know-how – Agents are pre-trained on industry terms, typologies, and skills.
- Internal policies and procedures – Sensa agents are then trained on your company information, ensuring that they align to your processes and risk appetite.
- Continuous learning – Sensa agents adapt and improve over time using your organization’s data.
- Testing and quality assurance – Agents continue to improve via self-learning and human feedback.
- Strong and robust model governance – Intuitive UI to manage and assess agent performance
.
Responsible AI is also at the heart of development with each SymphonyAI Sensa Agent being designed, trained, and maintained according to these core principles.
Introducing the Sensa agents
The first three Sensa agents have just been released. Here is a little more info on each of them:
Sensa Summary Agent
The Sensa Summary Agent provides detailed investigation overviews. As the name suggests, it summarizes internal data such as KYC, account information, and risk profile for any investigator to immediately understand alert context. Benefits include significant time savings, better investigator decision-making, and using data more effectively by breaking down silos.
How the Sensa Summary Agent works
Benefits of the Sensa Summary Agent
Sensa Narrative Agent
The Sensa Narrative Agent creates consistent and professional Suspicious Activity Report (SAR) narratives in seconds. Able to incorporate human and Agent-identified material for inclusion in the final narrative, the agent includes all relevant documents and case artefacts to provide a complete overview. Benefits include streamlined reporting, enhanced standards, and improved consistency across reports.
How the Sensa Narrative Agent works
Benefits of the Sensa Narrative Agent
Sensa Web Research Agent
Sensa Web Research Agent automates comprehensive background checks via online searches. Connecting to third-party sources and internal databases, the Agent delivers a detailed report which includes sources and relevance scoring. The benefits are numerous and include faster processing of false positives, significant time savings for teams, and the exposure of hidden risk signals from multiple sources.
How the Sensa Web Research Agent works
Benefits of the Sensa Web Research Agent
Get in touch – learn more about agentic AI and Sensa Agents
These first three Sensa Agents will be followed by many more, revolutionizing financial crime investigation process.
Learn more about SymphonyAI‘s agentic AI and Sensa Agents
Agentic AI FAQs
Agentic AI and AI agents are not the same thing, although they are related. Agentic AI refers to broader AI systems with the capacity for goal-directed behavior and autonomous decision-making, while an AI agent is a software tool designed for a specific purpose that perceives its environment and takes action to achieve its aim efficiently with limited scope.
AI agents are not likely to completely replace Robotic Process Automation (RPA), but they can complement and enhance it. While RPA is designed for automating repetitive, rule-based tasks, AI agents can handle more complex decision-making processes and adapt to dynamic environments, thereby expanding the capabilities of automation solutions.
AI agents and generative AI are distinct, though they can overlap in functionality. AI agents focus on autonomous decision-making and interacting with environments to achieve goals, while generative AI specializes in creating new content, such as text, images, or music, based on learned patterns.
SymphonyAI’s agentic AI page is an exceptional resource to help you understand AI agents. Alternatively, read the agentic automation Q&A to gain a better understanding of AI agents in financial services.
