Due diligence in the deal-making workflow requires the analysis of various information sources in a compressed period of time. These source documents provide information on market viability, financial health, capability and experience of management teams, and technology competency/proficiency. The documents are in various formats (PDFs, docs, presentations, audio transcripts, etc.) and there is no standard in terms of format, structure, or content. However, the information embedded in these information sources is crucial in building the business case for potential investments. 

Generative AI is powerful in unlocking the latent value in unstructured and semi-structured documents. Using this powerful AI/ML tooling, Hitachi Digital Services was able to extract insights and generate actionable and usable analysis for the deal workflow teams. This unburdening of basic document review allowed deal workflow associates to assemble pitch decks faster but, more importantly, to ask deeper questions on the documents themselves and be provided specific and targeted insights backed by citations and high confidence levels.

The AWS Bedrock offering provided a stable and mature platform to build custom GenAI solutions on serverless and managed services AWS products and services. This allowed our team to quickly provide a solution that could be used but, just as importantly, an AI/ML solution that can be managed by a small IT team past the production launch. The full solution leveraged the following AWS services: 

• API Gateway (Okta SSO) 

• Lambda (langchain RAG workflows) 

• Aurora with pgvector (Vector DB) 

• CloudWatch & CloudTrail 

• DynamoDB (chat history and session state) 

Hitachi Digital Services’ solution was deployed to a pilot set of users in a production setting. Through testing via manual workflow, the firm expects to achieve 20-30% efficiency gains.

Key Callouts: 

• Streamlined document review 

• Enhanced due diligence 

• Generative AI insights