AI Startup · Agent Unfolded
Arva AI
AI Agents for Financial Crime Compliance and Critical Banking Operations
London, United Kingdom · New York, United States
Arva AI builds evaluated and auditable AI agents for banks and financial institutions. Its platform automates work across financial crime compliance, business onboarding, payment operations and lending, combining proprietary models with decision infrastructure designed for regulated environments.
Company overview
Arva AI is a financial technology and regulatory technology company developing AI agents for high-stakes banking decisions. Its initial focus was Know Your Business (KYB) verification, helping banks and fintech companies assess business customers, review documents and complete compliance checks.
The company has expanded its scope beyond onboarding into broader financial crime compliance and banking operations. Its agents are designed to research individuals and businesses, investigate screening alerts, review transactions and support decisions across multiple workflows.
Arva positions its technology around three requirements for regulated institutions: accuracy, explainability and auditability. The company says its agents are evaluated against expert-labelled datasets and adversarial cases before deployment, with evidence and reasoning recorded for each decision.
Founded in 2024, Arva is backed by Gradient, Google’s early-stage AI fund, and Y Combinator. The company operates from London and New York.
Products and technology
KYB and business verification: Automates business onboarding checks, ownership research, document review, data extraction and risk assessment. The system gathers information from business registries, websites and other sources to support verification.
Screening and financial crime compliance: Helps compliance teams investigate sanctions, watchlist and other screening alerts, identify relevant information and reduce time spent reviewing false positives.
Transaction monitoring: Supports the investigation of suspicious activity by bringing together relevant information and producing documented findings for compliance teams.
Arva Intel: The company’s research model for investigating individuals and businesses online. Arva says its internal testing showed Arva Intel outperforming leading general-purpose models on accuracy by 13%.
Arva Research: Launched in September 2026, this applied research lab develops proprietary models for banking decisions where accuracy is particularly important. The company says its first research outputs are already running in production at financial institutions, including a top-10 US bank.
AgentCore: Arva’s continual learning infrastructure, designed to adapt agents to each bank’s operating procedures, policies and decision requirements over time.
Banking operations and lending: Arva is extending its agents into payment operations, disputes and lending, with the goal of automating more back-office decisions.
Funding history
Arva AI announced a $20 million Series A on November 17, 2026. The round was led by Arthur Ventures, with continued participation from existing investors Y Combinator and Gradient.
| Round | Amount | Announced | Lead investor |
|---|---|---|---|
| Seed | $3 million | January 14, 2025 | Gradient |
| Series A | $20 million | November 17, 2026 | Arthur Ventures |
The January 2025 seed round was led by Gradient, Google’s early-stage AI fund. Participants included Y Combinator, Amino Capital, Olive Tree Capital and angel investors from fintech companies.
Based on these two publicly announced rounds, Arva has raised at least $23 million. Its valuation has not been publicly disclosed.
Founders and leadership
Rhim Shah · Co-Founder and CEO: Shah previously led the financial crime product team at Revolut Business, where he worked on products and processes connected to business verification and compliance. He also held product leadership roles at Jobandtalent and studied engineering at the University of Oxford.
Oli Wales · Co-Founder and CTO: Wales previously worked as a lead product engineer at Seal, building applied AI systems for regulated environments. He studied computer science at the University of Cambridge and leads Arva’s research and engineering work.
The founders’ experience in fintech product development and applied engineering informs the company’s focus on building AI systems that can operate within the controls and requirements of financial institutions.
Investors
Arva’s funding includes specialist AI and fintech investors, alongside an early-stage venture firm with a focus on the financial technology sector.
Arva’s January 2025 seed announcement named Gradient, Y Combinator, Amino Capital and Olive Tree Capital. Arthur Ventures led the November 2026 Series A, with Gradient and Y Combinator participating again.
Customers and traction
Arva says its agents are deployed with banks and fintech companies in the UK and US. In its November 2026 Series A announcement, the company identified a top-10 US bank, an S&P 500 payments institution and global fintech Wise among its customers.
The company has also named Keep in Canada and Tola in the US as early adopters of its business verification platform in its January 2025 funding announcement.
Arva says its agents can automate up to 92% of manual compliance review in relevant workflows. It has also reported that its systems are being used to handle critical decisions with human oversight, rather than only providing summaries or recommendations.
Traction disclosure: Customer deployments, automation rates and performance figures in this section are based on company announcements. The identities of the top-10 US bank and S&P 500 payments institution have not been publicly disclosed by Arva.
Security, explainability and governance
Arva presents explainability and auditability as core product requirements for deployment in regulated financial institutions. Its stated approach includes evaluating agents against expert-labelled datasets and adversarial test cases before release, then regression-testing changes.
The company says its agents provide reasoning, supporting evidence and confidence information for decisions. It also describes versioned models, logged decisions and documented methodologies intended to support bank model risk management and regulatory review.
Arva publishes information about its security and AI risk management programmes through its Security Center and AI Risk Management resources. Specific certification details should be checked against the company’s current security documentation.
Frequently asked questions
What does Arva AI do?
Arva AI develops AI agents for banks and financial institutions. Its platform automates work across financial crime compliance, business onboarding, payment operations and lending.
How much funding has Arva AI raised?
Arva AI has announced a $3 million seed round and a $20 million Series A, bringing its publicly announced funding to $23 million.
Who led Arva AI’s Series A?
Arthur Ventures led Arva AI’s $20 million Series A announced on November 17, 2026. Existing investors Gradient and Y Combinator also participated.
Who founded Arva AI?
Arva AI was founded in 2024 by Rhim Shah, who serves as CEO, and Oli Wales, who serves as CTO.
Who are Arva AI’s investors?
Its publicly named investors include Arthur Ventures, Gradient, Y Combinator, Amino Capital and Olive Tree Capital, alongside angel investors.
Who uses Arva AI?
Arva has named Wise, a top-10 US bank and an S&P 500 payments institution among its customers. The two large US institutions have not been identified publicly.
What is Arva AI’s valuation?
Arva AI’s valuation has not been publicly disclosed.
Corrections or updates? If you spot an error or have information that should be updated in this profile, please contact hello@agentunfolded.com.
← Explore the AI Startup Directory