OutSystems launches governed AI loan applications for banks

OutSystems has launched Agentic Loan Applications for consumer lending, aimed at banks seeking more controlled use of AI in loan processes.

The pre-built package combines customer-facing applications, software agents and rules-based workflows that banks can adapt to their own products, policies and existing technology. It sits on top of current banking systems rather than replacing them, positioning it as a way to modernise loan application journeys without removing institutional control from lending decisions.

Banks have moved quickly to test AI tools, but lending remains one of the harder areas to change because of regulatory scrutiny, audit requirements and the need for human oversight. Many consumer lending journeys still rely on long application forms, manual document collection, repeated data entry and multiple handoffs before a loan officer can review a case.

That puts pressure on lenders from two directions: they need to improve the borrower experience while cutting operational friction, and they must do so in a way that remains reproducible and auditable for risk and compliance teams.

OutSystems says the new offering addresses the document-heavy work around lending decisions by combining mobile and web applications with data models, governed agents and deterministic workflows. The system can capture and verify documents, run identity and sanctions screening, and assemble a completed application for transfer into a bank's existing loan origination platform.

Governance focus

Governance is central to the launch. OutSystems says the agents are grounded in an operational context that includes a bank's policies, systems and dependencies, with the aim of keeping automated actions tied to institutional rules and existing processes.

The agents are also tested before deployment through multiple evaluations covering areas such as relevance, accuracy and protection of personally identifiable information. Any change to a model, prompt or tool triggers re-testing to keep the agents within defined guardrails.

Model economics are also part of the pitch. OutSystems says it tested multiple models through Amazon Bedrock and pre-selected ones that, in its assessment, delivered suitable performance at up to 82% lower cost. Banks still configure the system around their own products, rules and existing banking technology, while retaining control over their data.

OutSystems linked the launch to a broader push into industry-specific AI systems. Rather than offering standalone assistants, it is focusing on pre-built systems for regulated operational processes, where institutions want automation that can be monitored and reviewed.

"Banks do not need more isolated agents. They need governed agentic systems that improve real customer journeys and can stand up to risk review," said Luis Blando, CPTO, OutSystems.

"OutSystems brings agents into auditable workflows so institutions can move beyond isolated pilots and apply AI to complex lending journeys with the oversight the industry requires," Blando said.

Banking use cases

Several financial institutions already use the platform in regulated decision-making, including KeyBank, Paragon Bank and Axos Bank. The latest launch is designed to extend that position into consumer lending workflows, where banks have been balancing demand for faster digital journeys with caution over AI use in regulated decisions.

For lenders, the practical issue is not only whether an AI tool can answer questions or collect documents, but whether every action can be traced and reviewed. That is especially important where customer data, sanctions checks, identity verification and credit processes are involved, because those steps sit close to compliance and risk controls.

OutSystems is also making related development tools available through its Banking Agents Kit, which provides components for building banking applications and agents on its platform. This suggests the company is trying to address both ready-made deployments and custom development by banks that want tighter alignment with internal systems.

KeyBank's Mike Reynolds framed the issue in governance terms. "Hope is not a governance strategy. As AI adoption accelerates, banks need full visibility into how agentic solutions are built, what data they're accessing, and how they're being used," said Mike Reynolds, Business Technology Executive, KeyBank.

"Striving for centralized governance layers provides the oversight, auditability, and controls necessary to innovate confidently while staying within regulatory guardrails," Reynolds said.