How Nvidia's Tech is Driving Agentic AI Adoption in FinServ
While some sectors have been slow to jump aboard the AI train, companies in financial services have often been first movers.
In recent years, the use of Gen AI in customer service has risen from 25% to 60% across the sector.
In 2025, agentic AI appears to be the word of the year and, lo and behold, banks and fintechs have been at the vanguard of the technology.
Customer service has once again been an area of focus for the sector, with several high profile banks deploying AI agents to great effect.
The technology enables organisations to automate time-intensive tasks including document processing and report generation, delivering significant cost savings and operational efficiency gains.
Nvidia has played a huge role in this growth, with its technologies providing the backbone for so many agentic AI operations across the global economy.
According to a survey that Jensen Huang's company conducted recently, more than 90% of executives in the finance sector reported that AI had had a positive impact on their organisation's bottom line.
The key applications for agentic AI in the finance sector
Customer experience and engagement represents the primary use case for Gen AI, with 60% of respondents identifying this as their top priority.
Businesses deploying AI have already observed customer experiences improve by 26% as automated systems handle routine inquiries and processing tasks.
AI agents can process customer inquiries and forms while scaling support to ensure 24/7 availability, enhancing customer satisfaction levels.
This automation allows employees to focus on higher-level, judgement-based cases rather than performing case intake, data analysis and documentation.
Cybersecurity applications have experienced the highest growth over the past year, with more than a third of respondents now assessing or investing in AI for fraud detection and security purposes.
AI agents monitor transaction patterns in real time, learn from emerging fraud types and take immediate action by alerting compliance teams or freezing suspicious accounts without human intervention.
Putting the agents to work
BlackRock has integrated advanced AI capabilities into its Aladdin platform through Aladdin Copilot, using a federated development model where different teams can work on AI agents independently whilst building on a common foundation.
The investment management firm's central AI team established a standardised communication system and plug-in registry, allowing developers and data scientists to create and deploy AI agents tailored to their specific areas.
Capital One operates Chat Concierge, a multi-agent conversational AI assistant designed to enhance the automotive-buying experience for consumers.
The system provides 24/7 access to agents that deliver real-time information and take action based on user requests, including comparing vehicles and scheduling test drives or appointments.
Dutch bank bunq's Gen AI platform, Finn, now handles over 90% of all users' support tickets through its in-app chatbot functionality.
RBC's global research platform, Aiden, uses internal agents to automatically perform analysis when companies covered by RBC Capital Markets release SEC filings.
The technical infrastructure
Nvidia's software stack underpins these implementations through several key components.
The Nvidia Llama Nemotron family of large language models brings reasoning capabilities to AI assistants, enabling natural, humanlike interactions.
Nvidia NIM microservices provide industry-standard application programming interfaces for integration into AI applications and development frameworks.
Nvidia NeMo Retriever microservices enable the ingestion, embedding and understanding of relevant data sources, helping ensure AI agent responses remain relevant, accurate and context-aware.
The Nvidia NeMo Agent toolkit enables building, profiling and optimising AI agents through unified monitoring and detailed workflow profiling tools.
Nvidia NeMo Guardrails are implemented to ensure conversations with AI assistants remain safe and on topic, protecting brand values and maintaining customer trust.
These technical foundations support multimodal capabilities that can process queries combining text and images, making search processes more versatile for users.
The systems also manage both structured and unstructured data, crucial for processing the large volumes of complex financial information that institutions handle daily.
