Arva AI has raised $20 million in Series A funding to expand artificial intelligence models and agents designed for regulated banking and financial technology operations. Arthur Ventures led the round, with continued participation from Y Combinator and Gradient Ventures.
The company plans to prioritize investment in Arva Research, its internal model-development organization, while expanding its technology across banking back-office functions ranging from financial crime compliance to lending.
The financing follows several large deployments, including implementations at a top-10 U.S. bank, an S&P 500 payments company, and Wise.
Building Models For Explainable Financial Decisions
Banks make large numbers of decisions involving customer eligibility, financial crime risks, lending, transaction monitoring, and regulatory compliance.
Many of these decisions still require manual review because financial institutions must be able to explain their conclusions to auditors and regulators. General-purpose AI models may produce inconsistent or unsupported answers unless people verify the results.
Arva is developing proprietary models for decision stages where accuracy, evidence, and auditability are particularly important.
The company’s first research model, Arva Intel, investigates businesses and individuals using online information. In Arva’s internal testing, the model produced accuracy results 13% better than leading general-purpose alternatives.
Because this result comes from company testing, its broader performance will depend on the evaluation methodology, datasets, use cases, and models used for comparison.
Arva Intel operates alongside AgentCore, the company’s continual-learning infrastructure. AgentCore adapts the system to an individual bank’s policies, historical decisions, and operating processes.
The combination is intended to give banks AI agents that can perform research and operational tasks while producing results that institutions can review and explain.
Funding Follows Arva Research Launch
Arva launched Arva Research in September 2026. The new funding will support continued development of specialized models for banking operations.
The company’s strategy reflects a belief that financial institutions will require domain-specific systems rather than relying solely on general-purpose models. A banking agent must understand regulatory expectations, follow institutional policies, and maintain a record of how it reached a conclusion.
Boston Consulting Group has estimated that AI agents could reduce banking costs by between 30% and 40% by 2030. Achieving those savings would require institutions to move beyond pilots and allow AI systems to complete consequential work reliably.
Arva was founded in 2024 by CEO Rhim Shah and Chief Technology Officer Oliver Wales. Shah previously led financial crime products at Revolut Business, while Wales was formerly chief product engineer at Seal.
Their backgrounds combine regulated financial operations with AI product engineering. Arva is initially focused on financial crime compliance and lending, but it intends to extend its models and agents across additional back-office processes.
KEY QUOTES:
This year our customers started handing real critical decisions to Arva. How far that goes depends on accuracy, which is why we built Arva Research and why it’s where this funding goes first. Banks still run most of their back office by hand. We’re building the models that let them hand that work to agents and trust the result.
Rhim Shah, Co-Founder and CEO of Arva AI
Banks need AI they can trust with their most consequential decisions. Arva has built a rare combination of deep domain expertise, exceptional technical capability, and a clear commitment to accuracy and auditability. We’re excited to partner with Rhim and the team as they build the models and infrastructure for critical banking operations.
Patrick Meenan, General Partner at Arthur Ventures

