1. Case studies
C
a
s
e
s
t
u
d
i
e
s

FOREWORD

ln the dynamic realm of Enterprise Artificial lntelligence, the pursuit of innovation remains unyielding. As organizations around the world harness the might of data, Al, and various applications, the key to success lies in an infrastructure that is not only adaptable and scalable, but also impervious to threats.

Our Enterprise Al solutions encapsulate these pivotal qualities, equipping us with the capabilities to deliver stunning results for our clients.

LTlMindtree is thrilled to present you with a series of compelling case studies that narrate our collaborative journey in the realm of Enterprise Al. These studies offer a visual representation of how we empower our clients to unlock the immense possibilities of Generative Al technology.

But these case studies are more than mere stories. They are tangible proof of our unwavering commitment to assisting organizations in expediting their business growth exponentially.

We invite you to delve into these intriguing narratives and witness how we foster business augmentation with Al.

Happy delving!
Let’s explore Infinite Possibilities.
01
ENHANCED PRODUCTIVITY WITH
CONTENT GENERATION & SEARCH
02
CRITICAL DATA EXTRACTION FOR CONTRACTUAL DOCUMENTS USING GEN AI
03
FASTER SOP ACCESS AND SUMMARIZATION OF ANSWERS FROM MULTIPLE SOPs
04
INFRA AND APP AUTOMATION TO ENHANCE SUCCESS RATE WHILE REDUCING HUMAN ERROR
05
ENHANCED PRODUCTIVITY AND ACCURACY FOR RISK MONITORING & GOVERNANCE
06
NOTAM AI PROCESSOR TO REDUCE EFFORTS OF FLIGHT CREW

AI IN EVERYTHING

ENHANCED PRODUCTIVITY
WITH CONTENT GENERATION & SEARCH

Our client is an international financial institution dedicated to providing financial and technical assistance to developing countries worldwide. It is the largest and best-known development bank in the world and an observer at the United Nations Development Group.

CHALLENGES
  • Need for domain-specific, configurable knowledge bots for content search
  • Enabling multi-conversations/ interactions and enforcing access control
  • Operational staff spent significant time in finding accurate responses
  • Managing and retrieving relevant information from large volumes of documents is cumbersome
  • Often staff couldn’t find results even after days of searching
SOLUTIONS
  • Implemented a Gen AI-based Chatbot to search through structured and unstructured data
  • Ensured embedding from the data and stored them in a vector database
  • Used RAG-based contextualized search to retrieve accurate responses
  • Administrative role responsible for configuring chunking strategies, designing prompts, and performing prompt engineering
  • The end user references documents and receives the final generated response document
TOOLS & TECHNOLOGIES
  • Cosmos DB
  • Azure AI Search
  • Document Intelligence
  • GPT 4.0
  • Llama 2.0
  • Azure Blob Storage
BENEFITS

0%

Improved productivity of operational staff

0%

Enhanced accuracy in responses

Gen AI-based Chatbot to search data

AI IN EVERYTHING

CRITICAL DATA EXTRACTION FOR
CONTRACTUAL DOCUMENTS USING GEN AI

The client, a Fortune 500 company, operates in aerospace, building automation, industrial automation, and energy and sustainability solutions (ESS), addressing complex challenges in automation, aviation, and energy transition.

CHALLENGES
  • Create knowledge pools for critical data points from unstructured contract files on procurement and sales for units like aero, building automation, industrial automation, business documents
  • Retrieving relevant information from large volumes of documents is cumbersome and time consuming
SOLUTIONS
  • Generated high-quality, contextually relevant contractual content by leveraging advanced AI technologies and existing knowledge
  • Content information included 28 data points like Contracting Party, Business Conditions, Sales Price and Conditions, Timelines and Cost
  • Utilized Azure OpenAI services and other Azure technologies to deliver high-quality, contextually relevant content
  • Administrative role responsible for configuring chunking strategies, designing prompts, and performing prompt engineering
TOOLS & TECHNOLOGIES
  • Azure Open AI GPT 3.5 Turbo/ GPT 4/GPT 4o Mini
  • Prompt Engineering, Reranking, Embeddings
  • Azure services
  • Blob storage
  • Postgres DB
  • AI search
  • AI Document Intelligence
  • OpenAI GPT models
BENEFITS

0%

Improved productivity of operational staff

USD 264k per year

Annual productivity benefit

100-320 hrs/month

Reduction in contractual execution and validation time

Process

Enhancement from labor-intensive process to automated process

EVERYTHING FOR AI

FASTER SOP ACCESS AND
SUMMARIZATION OF ANSWERS FROM MULTIPLE SOPs

Our client is a privately owned company based in Vancouver, Washington, United States, that provides high-quality veterinary care to over 3 Million pets, including more than 2 Million Optimum Wellness Plan clients.

CHALLENGES
  • Engineers struggle to locate pertinent SOPs for issue troubleshooting amid a vast number of documents, causing inefficiencies
  • Managing over 3,000 servers is challenging, necessitating quick access to relevant SOPs for efficient troubleshooting
  • Handling more than 8,000 tickets requires streamlined access to SOPs to ensure timely and effective issue resolution
  • Navigating multiple tools (ServiceNow, SolarWinds, Cisco Meraki, SCOM) complicates the process of finding and utilizing the right SOPs
  • Reading and comprehending large documents significantly hampers productivity
SOLUTIONS
  • Efficiently searches 120+ SOPs, delivering concise and highly relevant answers
  • Meticulously gathered all necessary SOPs from the Delivery Unit for thorough coverage
  • Expertly categorized SOPs and created a highly efficient vector store
  • Seamlessly deployed the solution and conducted rigorous User Acceptance Testing (UAT)
  • Handles a wide range of queries, offering swift and reliable support to users
  • Delivers precise summaries from an extensive repository of SOPs with remarkable efficiency
TOOLS & TECHNOLOGIES
  • ServiceNow, SolarWinds, Cisco Meraki, SCOM
  • Python, Graph DB, Graph RAG
BENEFITS

Guidance for questions without SOPs

Faster SOP access

SOP viewing option from Chatbot

Summarization of relevant answers from multiple SOP

EVERYTHING FOR AI

INFRA AND APPS AUTOMATION TO
ENHANCE SUCCESS RATE WHILE
REDUCING HUMAN ERROR

The client is a US-based mass media entertainment company operating across six business units: Licensing, Broadcasting, Publishing, Streaming, and Television. Its activities include creating premium content and experiences for audiences worldwide.

CHALLENGES
  • Complete automation majorly around monitoring and request-based manual task
  • Standardize the process followed, create SN tickets for support and integration
  • Notification to stakeholders on required channel using a standard template
  • Centralized dashboard to track all automated solutions
  • Reduced bandwidth to be spent on recurring tasks
  • Segregation of SN tickets created via automation
  • Using a common and central repository to sort relevant data and files
SOLUTIONS
  • Scheduled, tool-based triggered, and UI-based solution deployed for different set of requirements following defined standards
  • Notification enabled on mailbox and SLACK as per requirement
  • Standard template followed for all tickets being created and updates made to required platform to fetch and populate execution count easily
  • Ensuring code integrity by having a standard repo on GitHub and security by using key for all sorts of creds
TOOLS & TECHNOLOGIES
  • 300+ Windows and LINUX Server
  • Orchestrator ST2
  • ITSM SN
  • Monitoring NR
  • Intelligent tool BP
  • 5+ Flavors of DB – MSSQL
  • MYSQL
  • MONGODB
  • POSTGRESQL
  • ORACLE
BENEFITS

0%

Reduction in human error as automation eliminates human intervention

0%

Reduction in time invested by the Team

0%

Success rate

0%

Efficiency

AI FOR EVERYONE

ENHANCED PRODUCTIVITY AND
ACCURACY FOR RISK MONITORING
& GOVERNANCE

The client is a major US-based bank, renowned for its extensive global operations and diverse client base. The bank has played a leading role in establishing important market intermediaries such as depositories, credit bureau, clearing and payment institutions.

CHALLENGES
  • Inconsistency and duplication of activities and controls across LOBs
  • Control inadequacy causing high ops risk & regulatory fines (recent USD 135 Million)
  • Inability to provide timely responses to regulators and lack of rational justifications
  • Challenges with lineage and traceability from MDRM to business process & data
SOLUTIONS
  • Standardized and discovered existing ARCM (Activities, Risk Compliance Management) inventory data through control descriptions
  • De-duplication of control data over 1 Million cross-LOB controls
  • Leveraged Gen AI to generate ARCM 4D descriptions:
    • − Option 1: Improved existing ARCM descriptions
    • − Option 2: Created new ARCM descriptions
  • Rationalized ARCM construct with lineage through parent-child ARCMs
  • Established unified risk & controls operating model and governance across LOBs
TOOLS & TECHNOLOGIES
  • Graph DB
  • GPT 3.5 for ARCM generation
  • DistilBERT for Rationalization
BENEFITS

0%

Improved productivity with risk controllers

Improved overall governance

Reduced regulatory finance expenditure

AI FOR EVERYONE

NOTAM AI PROCESSOR TO
REDUCE EFFORTS OF FLIGHT CREW

The client is a top specialist in air transport communications and IT, providing business solutions for airlines, airports, GDS, governments, and other customers globally.

CHALLENGES
  • NOTAMS (Notice to Airmen) are short, non-standardized and text-based instructions sent to flight crew and pilots from ATCs and airports before a flight starts
  • The flight crew must go through thousands of NOTAMs before a flight, filtering relevant and critical items and acting upon them accordingly
  • Extracting critical information from unstructured and non-standardized free text is time-consuming and labor-intensive
SOLUTIONS
  • Classify and tag the NOTAMs with Gen AI, based on their relevance and the criticality of alerts by airport, runway hazards, taxiway, ATC, and military ops
  • Harness open-sourced LLMs to extract and summarize NOTAM information
  • Transform the extracted information into a structured format to enable search and retrieval
  • Feedback loop to improve the accuracy of the pipeline over time
  • Optimized on-premise multi-GPU inferencing
  • Finetuning of Llama using LORA and PEFT techniques
TOOLS & TECHNOLOGIES
  • Meta Llama 3.2 and 3.1 and Azure Serverless Llama instances
  • Ollama and vLLM for LLM serving Docker
  • AKS
  • ACS
  • NVIDIA GPUs
  • Finetuning with LORA and PEFT
BENEFITS

0%

Reduction in flight crew efforts

< 1 sec per NOTAM processing

Reduced flight crew analysis time for NOTAMs from hours to minutes

© Copyright 2025 LTIMindtree Ltd. All rights reserved.