Arva AI Launches Dedicated Research Lab to Build the Models That Move Banking's Highest-Risk Decisions Beyond Human-in-the-Loop

The Lab's first outputs, proprietary decisioning models and AgentCore, are already used by global financial institutions, including a top-10 US bank

NEW YORK--(BUSINESS WIRE)--Arva AI, backed by Y Combinator and Google's AI Fund, announced the launch of its Research Lab, a research division building the proprietary models and infrastructure that let banks automate highest-risk decisions, already used by global financial institutions including a top-10 US bank.

Financial institutions rely on large teams to review cases across financial crime, fraud and other back-office functions, where errors cause financial loss, customer harm or regulatory breach. AI adoption has stayed human-in-the-loop: general-purpose models assist analysts but remain too error-prone and inconsistent for high-stakes automation.

"Banks keep humans in the loop because no AI has been accurate enough to remove them safely — that's the problem the Lab solves. Our models and AgentCore let us automate these decisions with the accuracy and control banks require, and this is just the start." — Rhim Shah, Founder and CEO, Arva AI

Over 5,000 hours of research, training and evaluation, the Lab built two things already in production at banks. Proprietary models for the highest-risk components of a decision — starting with enrichment and expanding to transaction analysis and evidence-based reasoning. And AgentCore, infrastructure turning analyst corrections, insights, and case outcomes into system improvements, each one backtested, evaluated and versioned before it reaches a live decision.

In independent evaluation against frontier models, Arva's first model, Arva Intel, used to research suspicious individuals and businesses online, outperformed general-purpose models by 13% on precision — benchmarked not on the final case outcome but on the accuracy of each component inside it.

The Lab starts with financial crime and fraud, where decision volumes are highest and errors most consequential, and will extend into payment exceptions, disputes and other customer-related investigations, and plans to publish its benchmark methodology and research at academic venues soon.

About Arva AI

Arva AI is a research company deploying autonomous AI decisioning for critical banking operations, combining proprietary models, adaptive decisioning agents and controlled continuous-learning infrastructure to improve the effectiveness, accuracy and repeatability of high-risk decisions.

Arva is backed by Google’s Gradient Ventures, Y Combinator, and other leading funds, with teams in New York and London.

Learn more at https://arva.ai/.