The transformational potential of generative AI have been the source of much speculation since the launch last November of ChatGPT, OpenAI’s Large Language Model-based AI chatbot. How might this technology be applied within the Capital Markets space in areas such as trading, market analysis, operations and regulatory compliance – and what challenges should be considered alongside the promised opportunities?

A natural language processing AI model designed to answer user input with human-like feedback, OpenAI’s ChatGPT was developed and pre-trained with large quantities of text data and then fine-tuned on particular tasks or niche focus areas. This model creates a type of deep learning neural network called a Large Language Model (LLM), which is then able to predict and generate context-relevant text strings that simulate the sentence structure of human language. 

Today’s AI industry is brimming with capital, talent, and is being driven by the world’s leading technology firms, many of whom have been developing their own chatbots and AI ecosystems in parallel to ChatGPT.

The use of AI is becoming more prominent in discussions about solving for data-driven decisions, given the potential for extracting better capital and labor cost efficiencies. With the rise of natural language processing (NLP) technology and its increasing availability and visibility, there is a growing interest in applying it to the area of Capital Markets. 

Source: Life Architect¹

Unlocking Capital Markets Business Potential with AI

As the scale of their operations grow, companies find it increasingly challenging to making timely and informed decisions – whether for the benefit of their internal stakeholders or external clients – given the increasing volumes of data that need to be surfaced and analyzed.

Trading:

• Use on trading platforms to supplement communications, operations, and inquiries between traders
• Provide 24/7 live trade support from upstream trading platforms to downstream depositories
• Deliver investment recommendations, advanced statistical analysis, and supplement technical or fundamental valuations
• Provide trade performance, reporting, and rebalancing.

Risk Analytics:

• Explore text data associated with financial risks, including market trends, regulatory changes, and economic indicators, to better identify these potential risks and create strategies for managing them
• Evaluate the design and effectiveness of internal controls and can help creditors or auditors identify deficiencies and assess the risk of material misstatement within financial statements.²

Client Support:

• Develop more sophisticated chatbots for a variety of purposes to support clients, such as onboarding new products and services, providing transaction updates, and processing trades where appropriate
• Providing 24-7 access to an AI ‘banker’ that can transfer customers to the relevant banking area in the event human assistance is required
• Offer real-time support to clients, answering frequently asked questions and guiding clients through complex processes.

Regulatory & Compliance:

• Propose pre-populated templates for legal contracts or trade agreements, identify gaps in regulatory reporting, and provide timely and efficient trade data querying capabilities
• Assist companies in meeting regulatory requirements by analyzing large volumes of data to identify potential compliance issues and provide guidance
• Process and interpret new financial markets legislation and weigh the costs and benefits of compliance implementation. 

Know Your Client (KYC):

• Analyze patterns in client behaviors or transactions and flag the potential for fraud or other malicious activities before they even occur³
• Search through internal client information, documents, and transactions to create a qualified client profile.

Market Analysis:

• Gather insights and generate reports on market trends, competitive landscape, and other relevant data
• Consume, analyze and present external market data and AI powered financial applications (i.e. BloombergGPT) to provide real time sentiment insights, named entity recognition and news classification.⁴

Banking & Transactions:

• Analyze performance trends in client transactions and recommend improvements to products and services for bankers
• Generate marketing content and sales proposals⁵
• Create transactional due diligence documentation for new debt/equity offerings, M&A, or private placements.

Operations & Technology:

• Suggest proven coding frameworks and lines to optimize portions of bank’s tech/engineering departments⁶
• Identify and resolve trade breaks and generate/allocate lifecycle and corporate action transaction events
• Generate documentation templates for client onboarding and legal risk and compliance
• Construct knowledge databases, find information quickly, and query data from internal documents
• Automate routine or highly manual tasks like recordkeeping, data entry, and invoice generation.⁷

Any potential benefits of AI must also be balanced by the potential risks. Data accuracy and quality in the broadest sense must be a particular area of focus. Capital Markets firms must take time to ensure that the data on which the AI is trained is vetted  via legal, risk, and compliance processes of some form. 

Within the industry, we have seen firms narrow ChatGPT’s training to include only the firm’s own proprietary data to generate responses based on only a more limited pieces of pre-vetted research. In this way, firms can be confident that the AI feedback is, in fact, accurate and in compliance with their respective policy guidelines. 

Despite the risks and hurdles of integration, it is important for Capital Markets firms to consider the business applications and potential benefits of using generative AI. Though it is not yet apparent the extent of these benefits to Capital Markets firms, Capco recommends assessing these applications and their impact on a firm’s success in an increasingly data-driven market landscape. 

If you have any questions or are considering AI chatbot technologies, please reach out to Trevor Williams and Randall Sawyer via our Contact Us form below. 

Contributors: Ervinas Janavicius, John Hamrick, Tyler Andringa

1. https://lifearchitect.ai/models/
2. https://www.linkedin.com/pulse/chat-gpt-future-auditing-risks-opportunities-andre-jacobs-ca-sa-
3. https://fintechnews.ch/aifintech/gpt-3-use-cases-in-banking-finance-and-fintech/58050/
4. https://www.bloomberg.com/company/press/bloomberggpt-50-billion-parameter-llm-tuned-finance/
5. https://southstatecorrespondent.com/banker-to-banker/technology/15-ways-we-are-using-chatgpt-in-banking/
6. https://www.capco.com/Intelligence/Capco-Intelligence/AI-Is-Redefining-Who-Can-Be-A-Software-Developer
7. https://www.financialexecutives.org/FEI-Daily/March-2023/A-Financial-Executives-Introduction-to-CHAT-GPT-3.aspx