Historically, most back-office business processes at financial services institutions have not been automated due to the inability of computers to effectively process unstructured data such as emails, PDFs, and images. Consequently, most legal, compliance, risk, and surveillance (LCRS) functions have relied on tedious, human-driven processes supported by basic IT workflows.
This obstacle has now been overcome. Business applications based on large-language models (LLMs) like OpenAI's GPT can perform basic reasoning tasks and process unstructured data.
Unfortunately, pinpointing the most advantageous opportunities for leveraging LLMs in LCRS business functions is not an easy task. Business customers (e.g. lawyers, compliance officers, risk managers) and technologists often employ incompatible language when discussing potential projects.
For instance, a compliance officer with deep expertise in regulation and policy may not understand how natural-language processing (NLP) capabilities, such as entity extraction and semantic search, are relevant to their work.
In our experience, most business customers only recognize opportunities for innovative technology when they see relevant examples based on their data.
This handbook is designed for analytics leaders at financial services companies tasked with brining innovative solutions to LCRS business problems. We have created use case examples based on real business problems and solutions we have deployed at the world's largest regulators and financial services companies.
These examples are generated using publicly available data and OpenAI's ChatGPT 4.0 model. We provide the prompts in the Appendix, so you can modify and run them yourself on this or other models.