Branch and Vault
Live
Banks

NatWest launches AI spending tool for RBS customers using Nvidia speech tech

NatWest has unveiled an AI-powered Spending Insights Tool for Royal Bank of Scotland customers, using Nvidia's speech recognition to better understand regional accents and natural language.

NatWest has unveiled an AI-powered Spending Insights Tool for Royal Bank of Scotland customers, using Nvidia's speech...

NatWest has unveiled an AI-powered Spending Insights Tool allowing Royal Bank of Scotland customers to explore finances through natural voice and text conversations. The tool, developed in partnership with Nvidia, uses advanced speech recognition to reduce word error rates by 20% across diverse UK accents.

The AI system allows customers to talk naturally about their spending, budgeting, and financial habits. It responds dynamically to follow-up questions, interruptions, and changes in topic. The experience combines conversational responses with interactive charts and visualisations, providing personalised insights based on a customer's own data.

Testing and rollout

The Spending Insights Tool is being tested with 75% of the NatWest Group's Royal Bank of Scotland customers. A wider Group rollout is planned after this initial testing phase is complete. The broader trial follows earlier internal testing of the underlying Nvidia artificial intelligence technology with the same customer base.

Nvidia partnership and speech recognition

NatWest partnered with Nvidia to improve the tool's speech recognition capabilities, focusing on understanding regional accents, pronunciations, and financial language across the UK. The collaboration uses Nvidia's Nemotron automatic speech recognition technology to convert spoken words into text.

The speech AI project began with in-depth testing involving 90 NatWest staffers from different UK regions. Staff accents included Glaswegian, Northern Irish, Welsh, Yorkshire, Mancunian, Midlands, Southern English, and non-British accents. Participants recorded around 3,000 realistic banking questions, both scripted and phrased in their own natural, colloquial language.

These recordings identified where speech-recognition models struggled. Merchant names, dates, and financial terminology were found to be particularly challenging. The system now mishears 20% fewer words, achieving an overall reduction in Word Error Rate across the accents tested. According to Nvidia's global head of banking, Aser Blanco, the real-time nature of speech recognition makes it challenging.

Industry context

Voice banking is becoming an increased focus for banks as they apply AI to conversational interfaces. Bahadir Yilmaz, chief analytics officer at ING, said conversational banking will be a fundamental experience people will look for in the next five years. ING's current conversational AI chatbot offers personalised information, problem solving and taking action, with the bank not far from enabling core capabilities like sending money and applying for products within regulatory boundaries.

Last month, digital challenger bank Zopa removed more manual tasks by offering customers the ability to bank through natural speech. Its feature, called Ask Zopa, allows customers to use voice commands to instruct an AI-powered personal digital banking assistant. The assistant understands conversational language and can manage tasks such as making transactions or setting savings goals.

Leadership and technology

Maja Pantic, chief AI research officer at NatWest Group, leads the bank's AI research. The Spending Insights Tool uses NatWest's proprietary Small Language Models to interpret the text generated by the speech recognition and to generate appropriate responses. The bank expanded this work using anonymised contact-centre transcripts to learn from a broader range of real-world banking language. A wider Group rollout is planned after testing.

Topics

#NatWest

Related coverage

More from Banks