How Generative AI Is Transforming the Financial Industry

Generative AI is rapidly spreading across finance—from document summarization and Q&A to customer service and legal document drafting.

However, adoption isn't easy for financial firms due to strict regulations around customer data protection, fairness, and market integrity. Even so, 85% of financial services companies are already using AI in some form, and plan to expand its application further within the next two years.

Key use cases include the following.

• Document summarization: Investment banking, research, and asset management sales teams need to read large volumes of unstructured documents and extract key points. Specialized summarization models like the Summarize API extract key points from single or multiple documents to provide summaries faithful to the source text, and Bloomberg has also adopted a similar summarization solution.
• Information extraction and Q&A: Contextual Answers uses NLP and machine learning to read context and find answers to questions, reducing hallucinations. Morgan Stanley has unveiled a generative AI solution that answers questions against its internal research library.
• Customer support: Chatbots that respond based solely on company database content can process inquiries quickly, boosting both operational efficiency and customer satisfaction. Capital One is experimenting with generative AI-based customer service automation.
• Call transcription and compliance: Items such as quantity, price, time, and execution status can be extracted from call recordings and reformatted into internal compliance formats. This leverages named entity recognition, and Goldman Sachs is running a POC to classify and reformat millions of legal documents.
• Legal document generation: Advanced LLMs can be used to draft proprietary legal documents for investment banks faster and with greater precision.

Ultimately, financial firms like JP Morgan, Bloomberg, and Morgan Stanley are already using generative AI for research, customer service, and document processing. However, since these are high-risk tasks, they must be paired with trusted partners, continuous human review, and ongoing monitoring.