RBC Capital Markets Brings Trusted AI to Bankers with S&P Global and Kensho

Through this collaboration, RBC integrated S&P Global’s trusted data directly into bankers’ AI workflows, resulting in fast, reliable results for clients.

Executive Summary

RBC Capital Markets equipped its front-line teams with AI to amplify human talent, provide faster and more insightful deliverables to clients and reinvest time saved to further strengthen its relationships. RBC put S&P Global's data and insights directly into Aiden, its Capital Markets AI platform, and into the daily workflows of its bankers, as efficiency and accuracy were non-negotiable. Working with S&P Global's Kensho Labs team, RBC integrated S&P Global financial data into Aiden and scaled to 8,000 employees, reaching over 98% accuracy on queries.   

Key Points

• RBC achieved over 98% accuracy on queries across its Capital Markets AI platform, Aiden, with the integration of S&P Global’s trusted data through the S&P Global AI Data Portal. 

• RBC’s bankers get fast, reliable answers grounded in accurate financial data, compressing hours of meeting prep into minutes. 

Better Tools for Bankers, Better Outcomes for Clients

RBC saw an opportunity to streamline information-gathering using AI, but only if the underlying data was trustworthy and the retrieval methods were failsafe. The faster and more reliably bankers can pull information together, the more time they have to focus on higher-impact work to better serve, understand and provide real value for their clients.  

The data has to be correct, the answers have to be auditable and the experience has to be fast and seamless. That high bar for rigorous quality and customer service is what led RBC to S&P Global and in particular, its Kensho team, which acts as the enterprise's AI innovation engine, helping customers unlock the value of S&P Global data across AI and agentic workflows. 

"Fast, trusted data in the hands of bankers leads to sharper analysis, stronger relationships and higher-value advice for clients. That's what our team set out to build — and why we partnered with S&P Global and Kensho."

-Bobby Grubert, Head of AI & Digital Innovation, RBC Capital Markets

Solving the Data Retrieval Challenge

As banks race to deploy agents and LLM-powered workflows, they often run into a similar challenge: models can hallucinate or pull from outdated sources. And in client-facing work, that disparity is disqualifying.  

Kensho is closing that gap on multiple fronts, building data retrieval solutions that make S&P Global's trusted data verifiable for LLMs across multiple access methods. For RBC’s use cases, the AI and the data it leveraged had to be reliable and trustworthy. Their bankers were asking structured questions that needed predictable, cited answers, rather than AI reasoning or generation. 

That standard pointed RBC to Deterministic Retrieval, an AI data retrieval solution was designed for exactly this problem. Offered through the S&P Global AI Data Portal, the solution integrates S&P Global’s trusted data into LLM workflows with the accuracy and auditability banks require: the same question returns the same cited answer, every time. Unlike general-purpose retrieval, the solution returns citation-backed answers from S&P Global's datasets, eliminating the hallucination risk that disqualifies most AI tools from client-facing work. For RBC, that meant the foundation was already in place and they could now focus on implementing the tools rather than inventing new ones. 

Deploying Deterministic Retrieval at scale required working within RBC’s rigorous compliance requirements and responsible AI principles, ensuring that the highest standards of accountability, fairness, privacy, security and transparency were upheld. The bank’s internal policies required a custom integration approach, while latency, accuracy and infrastructure requirements across thousands of users called for engineering depth beyond standard onboarding. RBC needed a collaborative partner that could match its velocity and engineer alongside its team.

Co-Engineering Through Kensho Labs

Kensho Labs, S&P Global’s team of forward-deployed AI engineers and experts, embedded with RBC, meeting with them several times each week. They helped configure Deterministic Retrieval to RBC's specific response time and output quality thresholds, resolved bugs and edge cases iteratively, and load tested at enterprise scale. Together, the teams built the custom deployment pattern that RBC's security and compliance posture required. 

The feedback flowed both ways. As RBC deployed Deterministic Retrieval, they pressure-tested it, surfacing real-world gaps and product improvements that Kensho engineers iterated on within days. That means a better deployment for RBC and a stronger product for every S&P Global customer who comes next. 

After one month, queries inside Aiden were returning answers with over 98% accuracy. More importantly, every month saved on infrastructure was a month earlier that bankers could put S&P Global data to work in client conversations.