CUSTOMER CASE STUDIES
Single end-to-end AI system, unlimited use cases
Featured
Global Pharmaceutical Company
Use Case:
Advanced analysis of complex manufacturing problem reports
150,000
Event reports
Millions
Of extracted entities
50ms
Root cause analysis
Problem
Pharmaceutical companies must meet stringent manufacturing and regulatory compliance requirements. Pre-existing methods to identify risks and reoccurrences in manufacturing problem reports are laborious (involving multiple people), non-scalable, subject to human bias, and prone to error. Traditional processes to detect event similarities and predict manufacturing deviation root causes can be quite time consuming and costly for the organization.
Solution
MimiLanguage and MimiData to automatically analyze text-rich historical problem reports to categorize deviations into major topics, predict event similarities, streamline the information retrieval process, and predict root cause and assignment of problems.
Intelligent
Analysis of complex manufacturing reports to identify key parameters required for evaluation of event reoccurrence
Automatic
Categorization and visualization of compliance events into major trends
Predictive
Root cause assignment of problem events improve QC processes and regulatory compliance
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiLanguage
- A.D.S.
- MimiData
- MimiCognition
DATA COMPONENTS
- Vault Direct Connect (VDC)
- Data Flow Controller
- Text Cleaning Service
- Data Transformation Service
- Meta Handler
APPLICATION COMPONENTS
- Entefy UI Framework
- Mimi Text Portal
- Document Source Manager
- Data2Graph
- Entefy Authentication Service
OTHER SYSTEM FEATURES
- Out-of-Band Data Management
- Event Sourced Architecture
- Common Authentication with Entefy Services
- Regulatory Drug Taxonomy
International Banking Institution
Use Case:
Cognitive processing and analysis of financial reports and news
10,000x
Faster than equivalent human effort
Thousands
Of topics and entities available
Problem
Understanding market trends and creating effective trade strategies is largely dependent on teams of analysts interpreting news, reports, social activity, and other text-rich sources of data. Keyword filtering systems often fail to properly contextualize results and public news sources are available to everyone, increasing the need for competitive differentiation.
Solution
Entefy Illuminate with Mimi AI summarization to read, summarize, and condense articles and other documents at more than 10,000x the speed of average human reading and comprehension. End result: giving analysts the power to consume exponentially more information and reduce time to trade.
Automatic
Extraction and connection of common topics, terms, and other detected properties across multiple data sources and feeds
Intelligent
Analysis of multi-domain written content from financial news and prepared reports
Scalable
Machine learning and compute infrastructure to accomplish the work of armies of human readers
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiLanguage
- MimiData
Data Components
- Vault Direct Connect (VDC)
- Data Flow Controller
- Text Cleaning Service
- Sync Adapter Library (SAL)
- Data Extraction Service
Application Components
- Entefy UI Framework
- Mimi Text Portal
- Document Source Manager
- News Management Service
- Entefy Authentication Service
Other System Features
- Out-of-Band Data Management
- Event Sourced Architecture
- Common Authentication with Entefy Services
International Equipment Manufacturer
Use Case:
Smart search, discovery, and recommendations for a vast technical knowledge base
2x
More accurate in ticket classification
100ms
Root cause assignment
$25 Million
Estimated annual savings potential
Problem
For equipment support and maintenance personnel, existing company knowledge bases are fragmented and contain limited search capabilities to access important technical documentation. This leads to an increase in escalation tickets created as well as time-to-resolution (TTR) for mission critical systems as knowledge bases continue to grow. Increasing TTR has direct financial consequences as machine downtime or other key issues can stall production or degrade production quality.
Solution
MimiData and MimiLanguage to reduce overall TTR by enabling AI-powered search, root cause classification, and problem diagnosis for equipment-related issues, leveraging the company’s vast technical knowledge bases of structured and unstructured data.
Predictive
Root cause assignment of reported escalations to isolate mission critical information and other technical documentation
Unified
Management of fragmented documentation and data assets sourced from multiple separate knowledge base systems
Improved
TTR with smart search and real-time, semantic root cause analysis
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiLanguage
- MimiData
- A.D.S.
Data Components
- Vault Direct Connect (VDC)
- Data Flow Controller
- Text Cleaning Service
- Sync Adapter Library (SAL)
- Data Extraction Service
Application Components
- Entefy UI Framework
- Mimi Text Portal
- Document Source Manager
- Entefy Authentication Service
Other System Features
- Out-of-Band Data Management
- Event Sourced Architecture
- Common Authentication with Entefy Services
Mass Media & Entertainment Conglomerate
Use Case:
AI video analysis for search and digital asset management
10,000
Identifiable objects
120ms
Scene retrieval time
Problem
Digital asset libraries containing large amounts of video content typically suffer from limited search capabilities, traditionally focused on filenames, dates, and other basic metadata. Enabling advanced search and discovery across large bodies of video can require massive amounts of human effort to accurately tag and label scenes, people, objects, actions and other attributes. This manual process is highly inefficient, error prone, and not scalable for today’s ever-growing content needs.
Solution
MimiVision to analyze large volumes of video content with automatic detection and extraction of valuable attributes such as everyday objects, standalone scenes, recognized faces (incl. celebrities), brand logos, colors, and more. This complex multi-attribute AI meta extraction and labeling helps power capabilities such as visual search, recommendations, object tracking, and content analytics. With the help of MimiLanguage, users can even ask for content by describing scenes and characters in full natural language.
Fully Integrated
And proprietary suite of computer vision models to enable advanced visual recognition and detection
Robust
AI tag storage & maintenance capable of powering diverse use cases from a central Entefy source
Improved
Content discovery enriched by more than 1 Billion visual feature vector parameters assessable in each frame
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiVision
- MimiLanguage
- MimiCognition
DATA COMPONENTS
- Vault Direct Connect (VDC)
- Data Flow Controller
- Media Management Service
- Data Extraction Service
- Meta Handler
APPLICATION COMPONENTS
- Entefy UI Framework
- Entefy Advanced Search Portal
- Media Source Manager
- Entefy Authentication Service
OTHER SYSTEM FEATURES
- Out-of-Band Data Management
- Event Sourced Architecture
- Common Authentication with Entefy Services
- Mimi SmartAgent Compatibility
Global Retailer & Manufacturer
Use Case:
Intelligent product cost and sourcing optimization
Thousands
Of products
30+ Million
Cost data points
Hundreds
Of suppliers
Problem
In today’s fast-moving and increasingly global economy, the ability to derive reliable data-driven insights is highly valuable but difficult to attain. In retail and manufacturing, the traditional processes to manage vendors, costs, and distribution are typically human intensive, costly, error prone, non-scalable, and not conducive to real-time analysis and forecasting. For companies with global footprints, the traditional analysis cycles can extend for months at a time which leads to tangible productivity and monetary loss each year.
Solution
MimiData and MimiVision to analyze and unify complex data from thousands of sources to ensure real-time optimization of pricing and sourcing for raw materials and components supporting thousands of products and involving hundreds of suppliers. This Entefy solution helps reduce key analysis cycles from months to seconds.
Intelligent
Analysis of costing information across brands, suppliers, regions, and more
Automatic
Monitoring and synchronization of BOM changes within products and components, virtually eliminating human error
Advanced
Visualization of product cost data, identifying outliers, highlighting patterns, and revealing trends in real time
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiVision
- MimiData
- MimiCognition
- Model Library
DATA COMPONENTS
- Vault Direct Connect (VDC)
- Data Flow Controller
- Sync Adapter Library (SAL)
- Data Extraction Service
- Data Cleaning Service
APPLICATION COMPONENTS
- Entefy UI Framework
- Entefy Smart Webforms
- Data Manager
- Data2Graph
- Entefy Authentication Service
OTHER SYSTEM FEATURES
- Variable User Permissions
- Event Sourced Architecture
- Common Authentication with Entefy Services
- Auto Multi-Source Data Capture
Financial Institution
Use Case:
Automated loan application processing and underwriting
15,000x
Improved efficiency over existing process
100%
Workflow automation (vs 30% in legacy automation system)
Transparent
Recommendations with explainable AI
Problem
For lenders, manual loan application review and approval processes are often human-intensive, not scalable, and prone to errors or inconsistencies. Further, traditional rules-based automation software is ill equipped to manage complexities in lending decision criteria and variability in submitted applications.
Solution
MimiCognition and MimiData to help evaluate loan applications and determine appropriate scores for decision-making. Speed up loan processing and increase loan approval accuracy for larger volume of applications without the need to add personnel.
Automated
Extraction, cleaning, and analysis of important lending data attributes across multiple systems
Explainable
AI that delivers loan applications underwriting decisions in real-time
Advanced
Data visualization and mapping of lending data to highlight outliers, patterns, and trends
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- A.D.S.
- MimiData
- MimiCognition
- Model Library
DATA COMPONENTS
- Vault Direct Connect (VDC)
- Data Flow Controller
- Data Extraction Service
- Data Transformation Service
- Meta Handler
APPLICATION COMPONENTS
- Entefy UI Framework
- Transaction Manager
- Data2Graph
- Mimi Workflow Visualization
- Entefy Authentication Service
OTHER SYSTEM FEATURES
- In-App Model Testing Library
- Event Sourced Architecture
- Common Authentication with Entefy Services
- Controlled Model Updates
- Data Obfuscation (incl. for PII)
Leading Semiconductor Services Company
Use Case:
Semiconductor die yield and wafer defect analysis
50
Distinct production processes
1,265
Defect clusters identified
Thousands
Of semiconductor yield maps
Problem
In the semiconductor manufacturing industry, maximizing yields and minimizing defects is critical to success both in terms of supply management and profitability. Many current methods used to model deviations and identify wafer defects during chip production rely heavily on human expert analysis of massive volumes of data. This data is generated by machines and software systems installed at various stages throughout the production line. Heavy reliance on expert human knowledge to interpret and isolate defect patterns is not only time consuming but costly and inefficient as well.
Solution
MimiData and MimiVision to analyze complex data reported by diverse sensors and tools at each stage of the semiconductor manufacturing process, isolating new deviation patterns and acceptable process thresholds for improved yield analysis. Multimodal machine learning to identify and extract combinatorial signals which may only be visible across multiple tooling stages and involve structured and unstructured (typically image) data.
Predictive
Assignment of defect regions and root cause for faulty processes
Multimodal
Signal intelligence to handle log and image data from multiple sensors and tools
Advanced
Data visualization and mapping of identified die defect clusters across tools and process stages
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiData
- MimiVision
- A.D.S.
- Model Library
Data Components
- Data Flow Controller
- Data Extraction Service
- Data Transformation Service
- Meta Handler
Application Components
- Entefy UI Framework
- Transaction Manager
- Data2Graph
- A.D.S. Bookkeeper
Other System Features
- In-App Model Testing
- Event Sourced Architecture
- Common Authentication with Entefy Services
Global Data Center Provider
Use Case:
AI root cause analysis for interconnection and network reliability issues
10 Million
System log records processed
Seconds
To complete months or years of manual effort
Problem
Performing root cause analysis (RCA) for network and interconnection reliability issues can be extremely time consuming and difficult due to the massive amount and diversity of system log data necessary to diagnose outages events or other service degradation.
Solution
A.D.S. and MimiData to automate time series data analysis, predicting pools of features to be used for root cause insights. Dramatically accelerate identification of patterns and trends with 20 Million observations evaluated and scored every second.
Automatic
Descriptive pooling and feature evaluation engine with configurable bin, pool, and model parameter options
Robust
Feature exploration and pre-model analysis capabilities that dramatically accelerate data science workflows, completing months or years of manual effort in just seconds.
Advanced
Reporting and data visualization of interconnection data to highlight outliers, patterns, and trends
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- A.D.S.
- MimiData
- Model Library
Data Components
- Data Flow Controller
- Descriptive Pooling Service Library
- Data Extraction Service
- Data Normalization Service
- Meta Handler
Application Components
- Entefy UI Framework
- Transaction Manager
- Data2Graph
- A.D.S. Bookkeeper
Other System Features
- In-App Model Testing
- Event Sourced Architecture
- Common Authentication with Entefy Services
Global Bank
Use Case:
AI-powered analysis of daily workforce activity data across international operations
1 Million
Tasks recorded daily
$100 Million
Annual savings identified
Hundreds
Of data repositories & systems
Problem
Given the breadth and scale of global operations across more than 50 countries, manual analysis of workforce data using traditional analytics methods has proven ineffective in baselining and assessing a global operating model. Despite significant investments in custom tools, dashboards, and data warehousing, those who depend on reliable, accurate, and consistent analysis of workforce productivity data are hindered in their ability to make correct decisions quickly and with proper context.
Solution
MimiData to intelligently analyzes large volumes of complex workforce data logs (incl. tasks, keystrokes, process records, etc.) from multiple disparate data sources to generate a unified operating model across employees, divisions, and offerings. Connect with the Mimi AI Global Operations Dashboard to visualize hidden insights and recommendations for operational improvement.
Unified
Management of fragmented data logs and records to provide a holistic view of key company operations
Intelligent
Detection of automation-eligible processes and workflows across service locations, divisions, and product groups
Automated
Workflow recommendations designed to improve worker productivity and improve service quality
Entefy components & technology used:
Entefy applications leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- A.D.S.
- MimiData
- Model Library
Data Components
- Data Flow Controller
- Workflow Registry Library
- Data Cleaning Service
- Meta Handler
Application Components
- Entefy UI Framework
- Transaction Manager
- Data2Graph
- A.D.S. Bookkeeper
System Features
- In-App Model Testing
- Geographical Reference Registry
- Event Sourced Architecture
- Common Authentication with Entefy Services
Cryptocurrency Investment Fund
Use Case:
AI-powered analysis of cryptocurrency activity and market news to forecast prices and manage portfolio risk
1 Million
Transactions per week
Live
Price updates
Instant
Price forecasting
Problem
The cryptocurrency market is subject to high volatility and is regularly impacted by non-standard events such as news, social posts, and “alt-coin” activity. Further, unlike traditional stock markets which localized in nature and allow trading only during specific hours in each day of the work week, cryptocurrencies are bought and sold 24/7/365—too much for any human individual to manage. This constant and highly volatile activity makes trading through manual orders or rule-based bot strategies ill-equipped to sustain high levels of portfolio returns.
Solution
MimiData and MimiLanguage to analyze real-time market activity, including cryptocurrency price feeds and key market news sources (incl. social networks, RSS feeds, media outlets, etc.), to support human traders in developing superior fund strategies that can generate higher sustained returns across a multi-coin portfolio. Expand price forecasting and market sentiment capabilities with traditional financial stocks to manage a multi-asset portfolio.
Multi-Interval
Price tracking using Mimi AI with instant model forecasts enabled on custom OHLC (Open-High-Low-Close) intervals
Integrated
Backtesting across multiple historical windows to automatically track trading models and strategies
Automated
De-noising and reconciliation of coin activity events against market news signals
Entefy components & technology used:
Entefy application leverage a robust library of configurable services and components which are pre-built for speed, scale, and consistency.
Highlights
MIMI AI
- Mimi AI Gateway
- MimiData
- MimiLanguage
- Mimi Smart Agent
- Model Library
Data Components
- Vault Direct Connect (VDC)
- Data Flow Controller
- OHLC Data Management Service
- News Management Service
- Sync Adapter Library (SAL)
Application Components
- Entefy UI Framework
- Transaction Manager
- Data2Graph
- Outbound Notification Handler
- Entefy Backtest Manager
System Features
- In-App Model Testing
- Active Learning for Models
- Event Sourced Architecture
- Common Authentication with Entefy Services
- Auto Multi-Source Data Capture
