Liberis debuts AI underwriting agent Ada for complex credit decisions

Embedded finance platform Liberis has launched Ada, an AI-powered underwriting agent aimed at accelerating credit decisions and improving fairness in complex funding scenarios. Built in-house and named after computing pioneer Ada Lovelace, the proprietary tool is being positioned as a force multiplier for underwriting teams navigating increasingly intricate risk assessments.

With 85% of Liberis’ financing decisions already automated, Ada is designed to augment—not replace—human underwriters by handling the more nuanced and time-consuming parts of the credit decisioning process. The system can reduce manual decision time by up to 50% for complex applications, the company says.

“Ada’s value goes far beyond just time savings,” said Rob Straathof, CEO of Liberis. “As humans, we all have biases and blind spots. Ada helps us uncover those. It flags what we might have missed and brings those risks to the surface, improves fairness, and helps us learn from past decisions.”

From data to decision

Ada processes data across a range of sources, including open banking feeds, Companies House filings, credit bureau records, and digital footprints. It then converts this information into risk signals—such as revenue trends or compliance flags—through a five-layered architecture:

• Data ingestion: standardizes and enriches raw data
• Feature engineering: distills structured risk indicators
• Decision support: generates confidence scores and recommended actions
• Human feedback loop: underwriters review, override, or confirm outputs
• Delivery systems: integrates insights into Liberis’ Vector platform

By embedding human feedback directly into the learning cycle, Liberis aims to fine-tune Ada’s business logic and model parameters over time.

Bias reduction and reassessment

One of the more novel aspects of Ada’s deployment is its use in post-hoc analysis. The system is being trained on past loans that resulted in defaults, as well as on rejected applications that may have been unfairly declined. According to Liberis, this retrospective analysis helps identify missed red flags and exclusionary patterns that might otherwise go unchallenged.

Straathof pointed to the mounting complexity facing underwriting teams as Liberis continues to scale: “As our business continues to grow, the complexity of the funding decisions our underwriting team has to make increases commensurately. Ada will help cut through that complexity and enable us to continue helping small businesses access the funding they require at the pace they need it.”

Roadmap toward multi-agent orchestration

The launch of Ada is the first step in a wider strategy to build a multi-agent AI platform capable of orchestrating various decision-making processes. The long-term aim is to enhance operational scalability without adding proportional headcount—a key consideration as Liberis targets new embedded finance partnerships.

The release also signals the company’s growing investment in AI infrastructure and data governance. By leveraging mature language models and deep data integrations, Liberis appears intent on reshaping how risk is assessed at scale—balancing speed with scrutiny.

No rollout timelines were disclosed for additional AI agents, but with Ada now live, the company is signaling a readiness to evolve beyond incremental automation toward systemic decision augmentation.

• Read more under:
• Announcements
• BankingTech