Smarsh and AWS expand AI compliance tools for banks

Smarsh has expanded its collaboration with Amazon Web Services to offer artificial intelligence compliance tools for regulated industries. Early use at financial institutions has cut compliance review workloads by about 77%.

The collaboration centres on Smarsh products built on Amazon Bedrock and Amazon Bedrock AgentCore, with the software also available through AWS Marketplace. The work has been underway since October 2025, as banks and other regulated organisations look for ways to use generative AI without falling foul of governance and oversight requirements.

Smarsh is focused on the compliance burden created by digital communications monitoring, a long-running issue for banks, investment firms and other regulated businesses that must keep records and detect risky behaviour across large volumes of messages, emails and other internal and external exchanges.

One global investment bank that uses the system reduced reviewer workload across millions of alerts by 77%, with less than a 2% loss in risk detection, according to Smarsh. The deployment also improved performance across both lexicon-based and machine learning models.

That result matters because compliance teams often face a trade-off between reducing false positives and maintaining enough scrutiny to satisfy regulators. Firms have been cautious about introducing generative AI into these workflows because they need systems that are explainable, auditable and defensible under regulatory examination.

Financial groups have increased spending on surveillance and archiving tools in recent years as regulators have stepped up scrutiny of record-keeping and off-channel communications. AI suppliers are now trying to position their products not just as automation tools, but as systems that can narrow review volumes while still identifying conduct and communications risks.

"Financial institutions do not need more AI experimentation-they need trusted, auditable outcomes," said Goutam Nadella, Chief Strategy Officer at Smarsh.

"The biggest barrier to enterprise AI adoption in financial services has been governance and compliance risk. Through our collaboration with AWS and our foundation on Amazon Bedrock and Amazon Bedrock AgentCore, we are removing that barrier, proving firms can achieve transformational efficiency gains while maintaining the high standards regulators expect," said Nadella.

Compliance focus

Smarsh said its tools are designed to turn archived communications data into a more active source of intelligence for compliance, legal and surveillance teams. It outlined separate applications for surveillance reviews, legal discovery and alert reduction for smaller firms.

In surveillance, review volumes can be cut by up to 77% with its Intelligent Agent system, while users can identify three to five times more real risks across multilingual communications, according to Smarsh. For legal teams, the company said its Discovery Agent can reduce investigation costs by up to 75% by accelerating early case assessment and reducing reliance on external legal advisers.

Smaller firms are also a target market. Small and mid-sized organisations can reduce compliance alert volumes by up to 60% with its Noise Reduction Agent, Smarsh said, describing it as a tool intended to help leaner compliance teams manage large numbers of irrelevant alerts.

The broader commercial backdrop is an AI market in which large cloud providers and specialist software vendors are increasingly working together to package models, data tools and governance processes for regulated sectors. AWS has been seeking to build out industry use cases on top of its generative AI services, while partners such as Smarsh bring sector-specific rules and workflows.

"At AWS, we believe regulated industries shouldn't have to choose between innovation and compliance," said Carol Potts, General Manager, US ISV sales at AWS.

"Smarsh has shown what's possible when you combine deep domain expertise with the scalability of AWS, delivering compliant AI solutions that financial institutions can deploy with confidence. We're proud to collaborate with customers who are setting a new standard for responsible AI adoption," said Potts.

Customer use

Smarsh also cited K1 Investment Management as an existing customer using its AI tools built on AWS. The private equity firm said the software had helped reduce false positives, a common problem in compliance monitoring systems that can flood teams with alerts requiring manual review.

"As an early adopter of Smarsh's AI Noise Reduction built on AWS Bedrock, we've cut false positives by 50% while gaining the regulatory defensibility our compliance team needs," said Jean Kisaka, Vice President and Chief Compliance Officer at K1 Investment Management.

"This helps our compliance team focus on what matters most, to help deliver smarter oversight with better resource allocation," said Kisaka.

Smarsh serves banks, brokerage firms, insurers, registered investment advisers and public sector bodies. The AWS collaboration is intended to support organisations moving from limited AI pilots to wider operational use. The software's availability through AWS Marketplace may also help customers who want to buy and deploy tools within existing cloud procurement arrangements.