FIS AML Investigations Agent Opens to All US Banks

FIS has opened its FIS AML Investigations Agent to every bank in the United States, including institutions that have never been FIS clients. The move, announced on October 15, 2026, turns an agent built for one compliance workflow into infrastructure available across American banking.

The FIS AML Investigations Agent was jointly announced in May 2026 with Anthropic, and it is designed to take on one of banking's most manual compliance tasks: investigating anti-money laundering alerts. According to the company's announcement on Business Wire, this is the first solution of its kind from a provider that already operates the core banking, payments and data systems that financial crime investigations depend on.

FIS developed the agent with support from Anthropic and design partner Amalgamated Bank, where the agent is now being deployed. According to Amalgamated Bank's analysis of its own investigation workflows against the agent's capabilities, the tool is projected to reduce Level 1 alert review time by more than 90 percent. Level 2 case investigation time could fall by approximately 98 percent, and preparing suspicious activity report narratives could take more than 80 percent less time.

Evidence assembly, not just alerts

The scale of the problem is what makes the agent's approach notable. United States financial institutions spend an estimated 35 to 40 billion dollars a year on anti-money laundering operations, according to figures from LexisNexis published in 2023. Research from AnChain.AI in 2025 found that analysts spend 40 to 60 percent of their investigation time assembling data before any analysis can begin.

Industry false-positive rates run between 90 and 95 percent, according to Datos Insights figures from 2023, and alert volumes keep climbing. According to FIS, the gap is not analytical capacity but the manual evidence assembly that consumes it. The agent attacks that bottleneck directly.

The FIS AML Investigations Agent maps a customer's type, historical activity and expected behavior against each alert before assembling evidence. The resulting summary is already calibrated to what is and is not normal for that customer. A seasonal spike in cash deposits reads differently from deposits deliberately kept just under federal reporting thresholds. A routine international transfer reads differently from an unexplained cross-border wire to a high-risk jurisdiction. The agent identifies those distinctions and explains its rationale to the analyst.

That framing matters. Stan Szwalbenest, head of enterprise fraud and financial crimes at FIS, said in the announcement that investigations stall not because analysts lack judgment but because evidence is scattered across systems and every alert gets treated the same way. According to PYMNTS' reporting on the announcement, the agent brings together the core, payments and data these investigations rely on, giving analysts a head start while keeping them in control of every decision.

Humans keep every decision

The FIS AML Investigations Agent is built on FIS Trust Assured, the company's framework for embedding governance, compliance and responsible AI into its products. Every output the agent produces is a recommendation. No case can be closed without explicit analyst confirmation.

Each statement the agent makes links to its source record, and each institution's risk logic, procedures and thresholds remain isolated from every other institution's. The controls are designed so that autonomy in gathering evidence stays separate from deciding outcomes.

This pattern — agents that act within hard governance boundaries — is becoming the template for serious deployments. Agent governance platforms that watch for misbehaving agents, such as the one built by Opsin, which raised $21M for AI agent security, address the same problem from the other side: making sure deployed agents stay inside the rules their owners set. Trust infrastructure for agents is also arriving through directories that route assistants to verified support destinations, as Zendesk's Open Service Directory does for AI assistants. The FIS agent belongs to this same wave: useful autonomy, but the human keeps the keys.

The first agent in FIS's banking roadmap

According to the announcement, the AML Investigations Agent is the first in FIS's roadmap of AI agents for banking. That sequencing is deliberate. Compliance investigation is one of the most labor-intensive workflows in banking, which makes it the sharpest test of whether agents can carry real operational load under regulatory scrutiny.

Opening the agent to non-clients is the other deliberate choice. According to FinTech Global's reporting, FIS made the agent commercially available to every US-based bank precisely because alert volumes keep rising across the industry, not just among its existing customers. A provider that runs core banking infrastructure selling its own agents to banks that run on rival cores is a sign of where the agent economy is heading: distribution follows capability, not incumbency.

Whether the projections hold in production will be the real story. The 90, 98 and 80 percent figures come from Amalgamated Bank's analysis of its workflows against the agent's expected abilities — projections, not measured results. But if the FIS AML Investigations Agent delivers even a fraction of that time back to investigators, the industry's costliest manual workflow will have a new baseline. The first bank is already live. The rest of the industry just got invited.