Goldman Sachs and Deutsche Bank Use Agentic AI Technology for Enhanced Trade Surveillance Systems

Banks are pioneering advanced artificial intelligence systems, known as agentic AI, that go far beyond traditional keyword scanning or preset rule-based alerts. Unlike conventional static monitoring tools, these new AI systems can reason dynamically through complex trading patterns in real time, highlighting suspicious behavior that demands human review.

Bloomberg recently reported that Goldman Sachs and Deutsche Bank are actively exploring these “agentic” AI solutions to enhance trading surveillance. Their objective is to bolster oversight by deploying software agents that can analyze live trading activities and detect anomalies suggesting potential misconduct.

Adaptive Surveillance Agents in Trading

Major banks typically utilize automated monitoring systems relying on fixed rules—such as triggering alerts when trades exceed predefined thresholds or deviate from benchmarks. Compliance teams manually investigate these alerts.

The challenge lies in handling the vast scale and complexity of modern markets. With enormous volumes of data across various asset classes, time zones, and trading platforms, static rules tend to generate many false alarms and often miss subtle, evolving forms of market abuse.

According to Bloomberg, agentic AI systems improve upon this by evaluating multiple data signals, comparing current trading behavior against historical trends, and identifying unusual combinations of actions—not merely matching trades to a checklist.

These intelligent tools are designed to augment human compliance teams, not replace them. Their main role is to surface complex cases requiring in-depth human oversight.

Deutsche Bank's Collaboration with Google Cloud

Bloomberg highlighted Deutsche Bank’s partnership with Google Cloud to develop AI agents capable of near real-time monitoring of vast datasets encompassing orders and executions. This system flags anomalies more efficiently than traditional surveillance.

The bank’s increasing investment in AI reflects a broader trend to apply generative AI and large language models beyond customer-facing chatbots, focusing instead on analyzing both structured and unstructured data streams related to trading conduct.

These AI agents excel at detecting “complex anomalies” by cross-examining trade relationships, timing, market conditions, and trader histories—rather than evaluating single transactions in isolation.

Human compliance officers remain responsible for investigating flagged alerts and deciding on subsequent actions.

Goldman Sachs’ Advancements in Agentic AI

Similarly, Goldman Sachs is advancing agentic AI integration in its surveillance frameworks. Investment in AI across trading and risk management now extends into compliance, deploying agents that operate with a level of autonomy to detect subtle or unconventional patterns that defy simple rule definitions.

For regulators, these tools offer earlier detection of potential market misconduct, reducing risks to market integrity and reputations. For banks, the value lies in managing the surge of alerts with greater precision while upholding rigorous oversight standards.

Understanding the Significance of Agentic AI

“Agentic AI” denotes AI systems that can take goal-directed actions autonomously, rather than merely responding passively to inputs. In practice, these agents choose what data to analyze next, integrate multiple signal sources, and escalate suspicious findings with minimal human prompting.

In trading, such AI might continuously monitor order flows, price changes, communication metadata, and historical trading behaviors to assess whether activities comply with expected norms.

It is crucial to note: these systems do not make final disciplinary decisions. Regulatory frameworks maintain human accountability, with AI serving to organize and highlight potential issues more effectively than static legacy tools.

A Broader Shift in Compliance Practices

The innovation lies in applying advanced generative AI architectures to internal controls and regulatory compliance functions. US and European regulators encourage firms to enhance detection of market abuse, though agentic AI is not yet mandated; firms must maintain robust monitoring systems regardless.

Adoption of such AI tools is expected to grow as they demonstrate value in meeting regulatory requirements. However, banks face challenges in ensuring models are explainable, mitigating bias, securing data, and maintaining rigorous audit trails to satisfy compliance standards.

Implications for the Financial Industry

Successful deployment of agentic surveillance technology could reshape compliance workflows. Staff may shift from processing numerous simple alerts to concentrating on evaluating complex, AI-identified cases.

This evolution won’t eliminate the need for human judgment but will refocus efforts toward deeper investigation where it matters most. In an era of accelerating data volume and speed, real-time pattern analysis beyond rule-based monitoring is increasingly essential.

(Photo by Markus Spiske)

See also:
Mastercard’s AI payment demo points to agent-led commerce

Interested in expanding your knowledge on AI and big data within the industry? Explore the AI & Big Data Expo, held across Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other leading technology conferences. Click here for further details.

AI News is brought to you by TechForge Media. Discover other upcoming enterprise technology events and webinars here.