Trovata Deploys Autonomous Treasury Agents on AWS Bedrock

From fragmented banking data to autonomous workflows

According to AWS, Trovata operates as an AI-native cash and liquidity management platform with direct connections to over 10,000 banks. Its cloud-based platform brings together cash reporting, forecasting, analysis and money movement in a single system.

Treasury teams often work across multiple bank portals, spreadsheets and legacy Treasury Management Systems. AWS says these arrangements can make reporting and reconciliation take hours or days, limiting teams’ ability to make real-time cash decisions.

Trovata launched what AWS describes as the first generative AI built specifically for treasury in 2023. Its latest updates move beyond delivering static insights, with persistent agents that run on schedules, triggers or rules. These agents can monitor activity, explain variances, detect anomalies and deliver updates without human intervention.

How Trovata built the agents on AWS

The system uses Amazon Bedrock for large language model inference and Amazon Bedrock AgentCore for agent orchestration, agentic workflows and role-based access control. Trovata also uses serverless AWS infrastructure for real-time execution at enterprise scale.

Trovata AI is embedded directly into the company’s treasury operating system rather than being added as a generic chatbot. Users can query multi-bank treasury data in natural language, including by asking, “What was our net cash position across all accounts yesterday?”

When a user submits a question, the AI retrieves data through Trovata’s verified API layer, analyzes it and presents the response within AgentCore’s secure, isolated environment. The architecture does not provide the AI with direct database access, according to AWS.

An enterprise RBAC layer keeps agents within the same permission boundaries as human users. AWS says agent queries and actions are logged for compliance review, while bank-branded deployment allows banks to offer the capabilities within their own applications.

Use cases and reported customer results

The main use cases include automated cash flow forecasting with scenario modeling, natural language queries, real-time anomaly detection and automated reconciliation summaries across hundreds of accounts. Persistent agents can also monitor, analyze and report on treasury activity continuously.

Trovata’s AI Agents Playbook contains ten agent patterns drawn from real customer deployments. AWS says most treasury teams have 5, 15 workflows that are ready for AI agents today.

Customer case studies cited by AWS report more than $1M in annualized cost savings, over 5 hours per day saved per accounting team and a 77 percent reduction in annual TMS costs. Treasury reporting has been reduced from hours or days to under a minute, according to the company information included in the AWS post.

What the deployment means for treasury teams

The architecture is designed to shift repetitive treasury work from manual operations to governed, continuous workflows. Teams can configure agents through Trovata’s Agent Builder-Configurator, while permissions, API controls and audit logs provide safeguards for enterprise use.

For banks, the bank-branded deployment model offers a way to provide AI-powered treasury features without building the underlying AI system themselves. For enterprises, the combination of multi-bank connectivity, natural language access and autonomous workflows is intended to make cash visibility and recurring treasury processes faster, while keeping agent activity within existing access controls.