AI in Financial Services

In financial services, thousands of AML alerts and KYC delays remain an everyday reality. Faced with regulatory oversight, fragmented data, and high precision requirements, general-purpose AI quickly shows its limits.

McKinsey found that 71% of financial services firms have adopted generative AI in at least one function, while BCG pointed out that 75% are still stuck at the pilot or POC stage. Evident Insights reported that banks' new AI use cases numbered 117 from January to June 2025, more than double the previous six months.

Some cases have delivered results. Bank of America's Erica handles over 58 million interactions per month, and more than 90% of employees use it in their daily work. The bank also rolled out AI coding tools to 17,000 developers, cutting costs and speeding up development.

The key barrier is trust. Deloitte named lack of trust as a major obstacle in finance and accounting workflows, and under Basel III, SR 11-7, EU AI Act, and SEC oversight, data lineage, human-in-the-loop controls, and auditable outputs are essential.

• What AI does well: automating repetitive knowledge work, compliance review, customer engagement, and leveraging internal data through RAG and context engineering
• What AI can't do alone: organizing fragmented sensitive data, guaranteeing accuracy without verification, deploying to production in a way that fits regulatory workflows, or replacing governance, security, and audit

Moving from pilot to production takes more than models alone. Orchestration, domain expertise, and governance must come together, and a partner-based approach also includes verified components, integration patterns, private deployment options, and structured knowledge transfer. Maestro supports planning, verification, and execution, while Jamba underpins long-context-based retrieval and reasoning. Ultimately, the next wave of financial AI will be won by those who secure trust while delivering measurable value.