Are you looking for smarter insights delivered directly to your inbox? Subscribe to our weekly newsletters for essential updates on enterprise AI, data, and security tailored for industry leaders.
The Evolution of Equity Trading Technology
Despite being a heavily regulated sector, equity trading has consistently led the way in technological advancements within financial services. However, many banks have approached the adoption of AI applications and agents with caution. For instance, TD Securities, the equity and securities trading division of TD Bank, launched its TD AI Virtual Assistant on July 8. This initiative targets front office institutional sales, trading, and research professionals to enhance their workflow management.
Insights from TD Securities
Dan Bosman, the Chief Information Officer at TD Securities, shared with VentureBeat that the primary objective of the virtual assistant is to provide front-office equity sales and traders with valuable client insights and research. “The first version of this began as a pilot, which we then subsequently scaled,” Bosman explained. “It’s really about accessing that equity research data that our analysts produce and making it available to the sales team in a user-friendly manner.”
Challenges in AI Scaling
The landscape of enterprise AI is evolving, influenced by power limitations, increasing token costs, and inference delays. Join our exclusive salon to learn how leading teams are transforming these challenges into strategic advantages. Discover how they are architecting efficient inference systems for real throughput gains and unlocking competitive ROI with sustainable AI solutions. Secure your spot to stay ahead: https://bit.ly/4mwGngO
The Natural Language Challenge
Bosman emphasized that working on a trading floor exposes users to specific jargon, making the context of their inquiries unique. Therefore, the AI assistant must communicate in a natural and intuitive manner while accessing insights generated by traders. The concept for the AI assistant originated from a member of the equity sales team. Fortunately, TD Bank has a platform named TD Invent, where employees can propose ideas which the innovation leadership team can evaluate responsibly.
“Someone from our equity research sales desk came in and said, ‘I’ve got this idea’ and brought it to TD Invent,” Bosman remarked. “What I appreciate most about this process is that when you create something truly remarkable, you don’t need to market it aggressively. People come to us expressing their needs and ideas, which is incredibly rewarding when we can integrate our investments in data, cloud, and infrastructure.”
Leveraging Advanced Technology
TD Securities developed the TD AI Virtual Assistant using OpenAI’s GPT models. Bosman noted that the bank collaborated with its technology teams and the Canadian AI company Layer 6, which TD Bank acquired in 2018, along with other strategic partnerships. This assistant connects with the bank’s cloud infrastructure, enabling access to internal research documents and market data, including 13F filings and historical equity data.
Bosman describes TD AI as a Knowledge Management System, which refers to its ability to retrieve and synthesize information into concise, context-aware summaries and insights, allowing sales teams to efficiently respond to client inquiries. Additionally, the TD AI Virtual Assistant grants users access to TD Bank’s foundation model, TD AI Prism, launched in June. This model is utilized across the entire bank, not just within TD Securities. During its launch, the bank indicated that TD AI Prism would enhance the predictive performance of TD Bank’s applications by processing 100 times more data, replacing its single-architecture models while ensuring customer data remains internal.
Overcoming Development Challenges
“The development presented unique challenges, as generative AI was relatively new to the organization at the project’s inception, necessitating careful governance and control management,” Bosman noted. “Despite this, the initiative successfully united diverse teams across the enterprise, fostering collaboration to deliver an innovative solution.” He highlighted one of the standout features: its text-to-SQL capability, which translates natural language prompts into SQL queries.
To train the assistant, Bosman mentioned that TD Securities implemented optimizations to streamline the process. “With patent-pending improvements in prompt engineering and dynamic few-shot example retrieval, we achieved the desired business performance through context learning,” he said. “Consequently, fine-tuning the underlying OpenAI model was not necessary for interacting with both unstructured and tabular datasets.”
A Broader Trend in Banking
TD Bank and TD Securities are not the only financial institutions exploring the transition from assistants to AI agents. BNY has informed VentureBeat that it has begun offering multi-agent solutions to its sales teams to help address customer inquiries, such as those concerning foreign currency support. Similarly, Wells Fargo has observed an uptick in the usage of its internal AI assistant.