Skip to main content

Summary 

Sonata Software helped a multinational travel and tourism company streamline the process of summarizing legal contracts, Statements of Work (SoWs), and other legal documents into a concise one-page format. Leveraging GenAI-based advanced document analysis and content extraction technologies, Sonata significantly reduced time and manual effort for the client’s 100+ global Legal & procurement team.

Client Overview

The client, a multinational travel and tourism company with a global presence and an annual revenue exceeding 10 billion euros, offers a comprehensive suite of services including travel booking, accommodation, cruises, and destination experiences.

Pressure Points

The customer’s legal & procurement team manually reviews extensive SOW documents and master service agreements (MSAs) to generate a concise two-page summary report form (SRF), this process is labour-intensive and time-consuming due to the volume and complexity of the documents. Other challenges include:

Lack of Summarization Solutions No suitable solution for summarizing legal contracts, SoWs, and related documents into a one-page format

High Manual Efforts Significant manual effort (1-2 person days) needed for populating SRF by reviewing various legal documents

Reliance on Legal Expertise Heavy dependency on legal subject matter experts for document summarization

Template Inflexibility Difficulty in customizing the SRF template for various procurement categories

Solutions

AWS Textract is used to extract content from PDFs, focusing on tables, which are converted to markdown format to enhance AI comprehension

Extracted content is segmented, and vectors are created using the AWS Bedrock Titan text embeddings model, stored in DynamoDB

During the result generation, vectors are generated for each description and cosine similarity is applied to compare these with the document vectors

The top 5 most relevant chunks are compiled and presented to the AI, utilizing few-shot prompting techniques to produce the final output.

The solution empowered users with the ability to preview, refine, and download the final SRF, ensuring adaptability and accuracy in document generation.

Results that Speak Volumes

By implementing Sonata's AI-driven SRF generation solution, the clients' global legal and procurement teams are expected to save over $500K annually due to decreased manual labor and reduced reliance on SMEs.

5X faster document summarization process

60% decrease in legal team’s manual effort

Reduced dependency on legal SMEs for summary document creation

Enabled simultaneous processing of multiple SoWs/contracts