ALL CASE STUDIES
AI Agent
Anomaly Detection
ChatLLM
Forecasting and Planning
Personalization AI
Predictive Modeling
PERSONALIZATION AI
A multi-product software company
The company used Abacus.AI to score and prioritize leads to efficiently allocate its sales and marketing resources
Problem
The company wanted to optimize marketing and sales resources by prioritizing the leads to target
Solution
Trained and deployed an Abacus.AI lead scoring model that helps the Company determine the top leads to target
Results
Lead scoring helped the company optimize its sales & marketing resources and achieve higher growth
7X improvement in conversions
Trained on ~10M leads
Handled 2Bn+ events
PERSONALIZATION AI
Large Audio Books Publisher
Abacus.AI reduced customer churn by identifying customers most likely to churn and engaging with them
Problem
The company wanted to reduce customer churn
Solution
The company used Abacus.AI to develop a customer churn prediction model to identify and engage with customers most likely to churn
Results
The company reduced its customer churn by 20%
Reduced customer churn by 20%
5M predictions / month
4X improvement in predictions
FORECASTING AND PLANNING
A popular tools company selling product on multiple platforms including Amazon and Home Depot
The company used Abacus.AI to optimize inventory to ensure high availability of stock while limiting excess inventory
Problem
The company wanted to optimally manage inventory
Solution
The company used Abacus.AI Demand forecasting model to predict the inventory requirement for various SKUs
Results
Resulted in predictions that are 72% more accurate than what the company achieved using its models
Forecasted on ~5,000 SKUs
Decreased the forecasted percentage error by 42%
Forecasted for 1 month to 12 month timeframes
PERSONALIZATION AI
A customer insights company
The company used Abacus.AI to generate personalized recommendations to its customers and increase revenue
Problem
The Company wanted to generate recommendations tailored to its customers
Solution
The Company used Abacus.AI personalized recommendations model
Results
Increased revenue by ~20% based on an A/B test
Increased revenue by 20%
Handles millions of requests/day
Under 20ms latency
ANOMALY DETECTION
A popular satellite television company
The company used Abacus.AI to build an Anomaly model to detect Quality of Service issues
Problem
Reduce churn by proactively reaching out to customers experiencing quality of service issues
Solution
Used Abacus.AI anomaly detection model to identify abnormalities in user session data
Results
Identified 12 types of service issues that could cause a customer to churn using 3 different methods
Processed user session data from 2.5M customers in real time
Calibrated on 2+ TB of data and over 1Bn events
Identified 12 types of service issues
PERSONALIZATION AI
Very large retail company
The company used Abacus.AI to generate cart recommendations using Abacus.AI’s vector store
Problem
The company wanted to recommend items to users based on what’s already in their cart in real time
Solution
Abacus.AI vector store was used to generate recommendations
Results
Generated recommendations under 20ms latency
Under 20ms latency
Scales up to 10,000+ Queries/Second
Handles millions of requests/day
CHATLLM
A financial services firm
The company used Abacus.AI to identify key pieces of information from customer call transcripts and classify them
Problem
The company wants to reduce costs by automating the categorization of calls based on the transcripts
Solution
Used Abacus.AI Named Entity Recognition to automate this process
Results
Achieved 99% accuracy in classifying, saving multiple hours of human effort
Classified call records into 15 classes
Achieved accuracy of 99%
Aggregated over 100 datasets
CHATLLM
Leading Health Insurance Provider
Used Abacus.AI ChatLLM to quickly answer complex insurance claims related queries from customers
Problem
Company wanted to automate and improve accurate handling of claims
Solution
Created Abacus.AI chatLLM with context on hundreds of documents across tens of distinct problem areas to instantly answer customer specific queries on any health policy
Results
Reduced claims processing time by 90%
Extracted and cleaned text from thousands documents
Processed 1m+ claims received per week
Saved millions of dollars by automating and reducing claims processing time by 90%
AI AGENT
Global Outsourced Manufacturing Company
Used Abacus.AI to organize and query thousands of bespoke customer-specific manufacturing contract terms with AI Agents
Problem
Company wanted to search, sort and index 1000s of detailed, bespoke contracts with unique terms across pricing, responsibilities, IP and timing
Solution
Created an Abacus.AI AI agent that can (i) answer a set list of questions about any contract and (ii) given a question, organize contracts by answers to that question (including running custom-calculations)
Results
Saved more than 120 people-hrs per week by automating manual document lookup to answer queries
Saved more than 120 people-hrs per week
Organized, and sorted contracts across 10 key metrics
Indexed and enabled search across 1000s of contracts
CHATLLM
National Personal Injury Attorney
The company used Abacus.AI to identify key pieces of information from customer call transcripts and classify them
Problem
Used Abacus.AI to automate the creation of claimant-specific demand letters through AI agent
Solution
Company wanted to automate and streamline the labor intensive process of creating claimant demand letters, including data entry and state judicial document review
Results
Achieved 99% accuracy in classifying, saving multiple hours of human effort
All sensitive data secured safely
Ingested thousands of demand letters, depositions and positions to optimize output
Saved $1M+ by reducing demand document creation time by 60%
AI AGENT
Global Vacation Rental Company
Used Abacus.AI to enable internal company search across all contracts
Problem
Company wanted to enable search for tens of thousands of documents stored in company cloud
Solution
Used Abacus.AI document retriever and AI agent to answer any question on extracted documents
Results
Increased revenue 20%+ through better matching customer needs to property
100,000+ documents
25,000+ properties
Thousands of different property owners
AI AGENT
National Property Insurer
Used Abacus.AI to create DataLLM for all customer insurance policies
Problem
Company wanted to be able to search all insurance policies by various data fields
Solution
Used Abacus.AI data LLM to index and enable search on all policies
Results
Saved millions of dollars in support agent time and improved customer satisfaction scores
Millions of documents processed
20 searchable fields
Hours saved in daily support time
PREDICTIVE MODELING
B2C Subscription Organization
The company used Abacus.AI to reduce customer attrition and increase revenue using a predictive model for identifying at-risk customers
Problem
Existing models were not accurately identifying the customers likely to churn, resulting in lost revenue
Solution
The company used Abacus.AI Customer Churn solution to develop a model to accurately identify, and engage with, customers most likely to churn
Results
The company reduced customer churn, and achieved higher growth and revenue
Deployed a model that makes 5m predictions/month
4x improvement in accuracy
Reduced customer churn by 20%
PREDICTIVE MODELING
Food & Beverage Consumer Goods Company
The company used Abacus.AI to optimize marketing efforts by targeting leads with high conversion potential, significantly improving efficiency and sales outcomes
Problem
The company wanted to optimize sales and marketing resources by prioritizing leads to target over large amounts of data
Solution
The company used Abacus. AI Predictive Lead Scoring solution to develop a model to accurately determine top leads to target
Results
The company improved sales and marketing efficiency and increased conversion rates
Improved conversion rates by 7x
Trained a model on 10M rows of data
80% accuracy
FORECASTING AND PLANNING
Consumer Retail & Manufacturing Company
The company used Abacus.AI to enhance inventory management and demand forecasting, reducing errors
Problem
The company wanted to forecast demand across 5000 SKUS to optimally manage inventory but had difficulty building accurate models
Solution
Used Demand Forecasting solution to predict the demand for each SKU
Results
Improved their models and predicting the demand more accurately
Built models were 72% more accurate than existing models
Decreased forecasting error by 42%
Predicted demands from 1-12 months in the future
PERSONALIZATION AI
Multinational Consumer Products Company
The company used Abacus.AI to increase e-commerce revenue by personalizing user experiences on their website
Problem
The company wanted to personalize the user experience on their website to increase revenue but many visitors had limited shopping histories
Solution
Used Personalized recommendation solution to build a streaming recommendation model
Results
Provided recommendations for all visitors, even those with limited histories
Increased overall orders by 10%
Over 10M page views
10K orders per day
CHATLLM
Household Consumer Goods Organization
The company used Abacus.AI to streamline customer service operations by automating the analysis of call transcripts
Problem
The company was spending too much time and money manually categorizing call transcripts to improve customer service
Solution
Used Text Classification and Extraction solution to build a model that could analyze the call transcripts
Results
Increased workplace efficiency, nearly eliminating manual reviews
Automated transcript reviews with 99% accuracy
Aggregated data from 100+ datasets
Classified transcripts into 15 classes
PREDICTIVE MODELING
Global eSignature Company
The company used Abacus.AI to minimize customer loss by implementing an advanced churn prediction system
Problem
Existing models were not accurately identifying the customers likely to churn
Solution
Used Customer Churn solution to accurately identify and engage with customers more likely to churn
Results
Reduced customer churn and increased revenue
Reduced customer churn by 20%
Deployed a model that makes 5m predictions/month
4x improvement in accuracy
PERSONALIZATION AI
Multinational Home Improvement Retail Company
The company used Abacus.AI to boost online sales and reduce product returns by integrating advanced real-time recommendation models
Problem
The company needed to increase conversion rates of online sales while also decreasing returns to boost revenue
Solution
Leveraged Recommendation solution for real-time recommendations and plug-and-play PyTorch & TensorFlow models
Results
Deployed real-time recommendation models that were moved into production and integrated with their website
6 different real-time models deployed
Sale conversion rates increased by 3%
Returns decreased by 10%
PREDICTIVE MODELING
Financial Services
The company used Abacus.AI to implement advanced fraud detection capabilities in wire transfers
Problem
A leading financial services company wanted to identify fraud in wire transfers
Solution
Used a supervised model to identify transactions that are fraudulent
Results
Increased fraud detection and improved existing models
Caught 85% of fraud
Improved accuracy of existing models by 30%
FORECASTING AND PLANNING
Consumer Retail & Manufacturing Company
The company used Abacus.AI to improve the accuracy of demand forecasts, streamlining operations and reducing monthly expenditures
Problem
The company wanted to improve forecast accuracy for 1m SKUs over a period of six months
Solution
Used Demand Forecasting Solution to predict demand for each SKU
Results
Improved their models and predicting the demand more accurately
Built and deployed a model within 2 weeks
Spends $15k less per month on AI
Increased accuracy by over 25% across all categories
PREDICTIVE MODELING
Personal Styling B2C Company
The company used Abacus.AI to increase customer satisfaction and revenue by using predictive modeling
Problem
The company wanted to reduce costs and increase revenue by sending clothing that was more likely to be kept by their >100k subscribers
Solution
Used Predictive Modeling solution to build a custom classification model to predict the likelihood of an item being kept
Results
The percentage of items kept by customers increased
Percentage of items kept increased by 30%
Revenue increased by 18%
Shipping cost decreased by 8%
PERSONALIZATION AI
Health Care
The company used Abacus.AI to enhance patient engagement and decision-making on their healthcare portal by deploying recommender systems
Problem
The company wanted to leverage their next best actions on their portal, but many of their patients had very little history
Solution
Used Recommender solutions to build high-performance models alongside heuristics
Results
The company deployed real-time recommendation models as part of their portal
Healthcare portal provides recommendations
In a B2B scenario, opportunity won conversion 10%
FORECASTING AND PLANNING
Health Care
The company used Abacus.AI to prioritize research investments more effectively by forecasting drug demand with new models
Problem
The company wanted to prioritize research investments with forecasted drug demand
Solution
Used Demand Forecasting solution to build models for all medical conditions related to their drug portfolio
Results
Improved forecast accuracy compared to econometric models
Forecasting NN 25% more accurate than other models