Generative AI Revolution:
Transforming Business with
GenAI-Infused Software
Prepared by Gregory Entin
Generative AI and
Current Tech
THE RESULT OF YEARS OF HARD WORK
• Dedicated research
• Experimentation
• Gradual improvements
History of GPT Large Language Models
GPT-1
Jan. 11, 2018
GPT-2
Feb. 14, 2019
GPT-3
May 28, 2020
ChatGPT
Nov. 30, 2022
GPT-4
Mar. 14, 2023
§ A subset of AI
§ Based on Neural Networks and Machine
Learning models
§ Trained on vast amounts of data
§ Can generate new content based on the learned
patterns
§ Convolutional Neural Networks (CNN) for
graphics
§ Recurrent Neural Networks (RNN) for text
Understanding Generative AI
§ Expert Systems
§ Data Extraction
§ Data Categorization
§ Automated Quality Control
§ Object Detection
§ Speech Recognition
§ Personalized Recommendations
§ Predictive Analytics
§ Autonomous Vehicles
§ Facial Recognition
§ Chatbots
§ Deep Learning
§ Robotics
§ Healthcare Diagnostics
History of AI Applications
Generative AI
On the Verge of Artificial
General Intelligence
§ CONTEXT: AGI remains a goal, and Generative AI
serves as a pivotal step towards achieving this
milestone
§ GENAI'S IMPACT: Leveraging deep learning, it
pushes technological boundaries, automating
tasks across domains
§ VERSATILITY OF GPT: GPT embodies the
adaptability, applicable to a wide array of tasks
§ DIFFERENTIATOR: GPT's versatility marks a
paradigm shift from traditional AI technologies
GPT – A Universal Tool
§ Marketing, copywriting, SMM
§ Complex data extraction and normalization
§ Quality control
§ Prioritizing and business decision-making
§ Ideation
§ Estimating and forecasting
§ Risk management
§ Education
§ Customer service
§ Content personalization
§ Recruitment and HR
§ Healthcare diagnostics
§ Legal document analysis
§ Predictive maintenance
§ Sentiment analysis
Generative AI in Business Today
GenAI
Case Studies
MATCHING HEALTH INSURANCE PLANS WITH
MEDICAL CODING FOR ELIGIBILITY AND COVERAGE
ESTIMATES
§ Health insurance plans – poorly structured text
§ Identifying policy eligibility and coverage
§ Automated calculation of:
§ Coverage amount
§ Deductible
§ Co-payment
§ Co-insurance
§ Out-of-pocket maximum
§ Claim amount
§ Reimbursement amount
Case Study: Healthcare Insurance Verification
PROVIDING AN AI TOOL FOR CORPORATE EMPLOYEES
§ Built into native corporate environment via
MS Teams integration
§ Corporate access control
§ Secure
§ Adherence to privacy policies
§ High reliability
§ Usage and cost control
§ Fully auditable
§ Available from any devices
Case Study: Corporate AI Chatbot
IMPLEMENTING AN ERP ADD-ON FOR EMPLOYEE
TIMESHEET RECORDS VALIDATION AND CONTROL
§ Automatic analysis of work time records
§ GPT-based, provides accuracy of 85% and more in
detection of incorrect records
§ Escalating and reporting to management
Case Study: AI Timesheet Records Control
CREATING PERSONALIZED LEARNING
RECOMMENDATIONS FOR PARENTS OF STUDENTS
§ Real-time recommendations based on student
learning data
§ Personalized learning paths, individual for every
student
§ Predictive analytics: predicting student
performance, identifying at-risk students
§ Learning analytics: offering insights to teachers
and administrators
Case Study: AI Recommendations for K12
Challenges of GenAI
Implementation
Challenges Overview
PROMPT ENGINEERING ROLE PRIVACY AND SECURITY
QUALITY ASSURANCE, STABILITY,
PERFORMANCE
GPT RESPONSES VS. MACHINE-
READABLE STRUCTURES
COST CONTROL AND
OPTIMIZATION
COMPLETELY NEW ROLE, NEVER EXISTED
§ Unique combination of skills
§ Need real experience
§ Hard to find on the market
Prompt Engineering Role
EXTREMELY IMPORTANT
§ Data encryption at rest and in transit
§ Access control
§ Data anonymization
§ Compliance with privacy laws
§ Incident response plan
§ AI model transparency and auditability
§ User consent and data rights
§ Limit the use of customer data for model training
Privacy and Security
NEW TECHNOLOGIES COME WITH RISKS
§ OpenAI platform availability and outages
§ Unpredicted model behavior changes
§ Unstable results based on input data
§ Degraded performance at peak hours
§ Robust testing protocols
§ Model validation and verification
§ Continuous monitoring and feedback loops
§ Version control and model management
§ Caching and failover
§ Performance profiling and optimization
Quality Assurance, Stability, Performance
Gen AI Costs Can be Significant
§ Transactional costs / per token
§ Cost profiling
§ Using different models (price differ x10, or x100)
§ Pre-processing with cheaper models
§ Custom models to run on-premise
§ Constant monitoring and improving
Cost Optimization and Control
HOW TO UNDERSTAND A ROBOT?
§ Ask for Structured Data Formats (JSON/XML, etc.)
§ Additional AI models (NER and others) for data
extraction
§ NLP Libraries (e.g. NLTK)
§ Built in quality control
GPT Responses vs. Machine-Readable Structures
Welcome to the
New Era of GenAI
VIRTUALLY EVERY SYSTEM WILL USE IT
COMPANIES NOT USING GENERATIVE AI
WILL BE LOSING IN COMPETITION
COMPANIES WILL NEED TO GAIN GENAI EXPERTISE
Future of Business with GenAI
1
2
3
LARGE VENDORS WILL OFFER NEW TOOLING
4
§ Plan for the rise of
competition
§ PoC can be done in weeks
or even days
§ All tools are out there
§ # Years in AI business
§ Own AI-powered solutions
§ Expertise in AI, prompt
engineering
§ Expertise in your domain
STRATEGICALLY THINK ACT QUICKLY
EVALUATE AND SELECT YOUR
AI PARTNER
Getting Your Business Ready for GenAI
Subscribe to Our Newsletter on LinkedIn
DON’T MISS THE
NEXT EDITION!
Thank You!
+1 (847) 559-0864
sales@velvetech.com
www.velvetech.com

GenAI Revolution: Transforming Business with GenAI-Infused Software

  • 1.
    Generative AI Revolution: TransformingBusiness with GenAI-Infused Software Prepared by Gregory Entin
  • 2.
  • 3.
    THE RESULT OFYEARS OF HARD WORK • Dedicated research • Experimentation • Gradual improvements History of GPT Large Language Models GPT-1 Jan. 11, 2018 GPT-2 Feb. 14, 2019 GPT-3 May 28, 2020 ChatGPT Nov. 30, 2022 GPT-4 Mar. 14, 2023
  • 4.
    § A subsetof AI § Based on Neural Networks and Machine Learning models § Trained on vast amounts of data § Can generate new content based on the learned patterns § Convolutional Neural Networks (CNN) for graphics § Recurrent Neural Networks (RNN) for text Understanding Generative AI
  • 5.
    § Expert Systems §Data Extraction § Data Categorization § Automated Quality Control § Object Detection § Speech Recognition § Personalized Recommendations § Predictive Analytics § Autonomous Vehicles § Facial Recognition § Chatbots § Deep Learning § Robotics § Healthcare Diagnostics History of AI Applications
  • 6.
    Generative AI On theVerge of Artificial General Intelligence
  • 7.
    § CONTEXT: AGIremains a goal, and Generative AI serves as a pivotal step towards achieving this milestone § GENAI'S IMPACT: Leveraging deep learning, it pushes technological boundaries, automating tasks across domains § VERSATILITY OF GPT: GPT embodies the adaptability, applicable to a wide array of tasks § DIFFERENTIATOR: GPT's versatility marks a paradigm shift from traditional AI technologies GPT – A Universal Tool
  • 8.
    § Marketing, copywriting,SMM § Complex data extraction and normalization § Quality control § Prioritizing and business decision-making § Ideation § Estimating and forecasting § Risk management § Education § Customer service § Content personalization § Recruitment and HR § Healthcare diagnostics § Legal document analysis § Predictive maintenance § Sentiment analysis Generative AI in Business Today
  • 9.
  • 10.
    MATCHING HEALTH INSURANCEPLANS WITH MEDICAL CODING FOR ELIGIBILITY AND COVERAGE ESTIMATES § Health insurance plans – poorly structured text § Identifying policy eligibility and coverage § Automated calculation of: § Coverage amount § Deductible § Co-payment § Co-insurance § Out-of-pocket maximum § Claim amount § Reimbursement amount Case Study: Healthcare Insurance Verification
  • 11.
    PROVIDING AN AITOOL FOR CORPORATE EMPLOYEES § Built into native corporate environment via MS Teams integration § Corporate access control § Secure § Adherence to privacy policies § High reliability § Usage and cost control § Fully auditable § Available from any devices Case Study: Corporate AI Chatbot
  • 12.
    IMPLEMENTING AN ERPADD-ON FOR EMPLOYEE TIMESHEET RECORDS VALIDATION AND CONTROL § Automatic analysis of work time records § GPT-based, provides accuracy of 85% and more in detection of incorrect records § Escalating and reporting to management Case Study: AI Timesheet Records Control
  • 13.
    CREATING PERSONALIZED LEARNING RECOMMENDATIONSFOR PARENTS OF STUDENTS § Real-time recommendations based on student learning data § Personalized learning paths, individual for every student § Predictive analytics: predicting student performance, identifying at-risk students § Learning analytics: offering insights to teachers and administrators Case Study: AI Recommendations for K12
  • 14.
  • 15.
    Challenges Overview PROMPT ENGINEERINGROLE PRIVACY AND SECURITY QUALITY ASSURANCE, STABILITY, PERFORMANCE GPT RESPONSES VS. MACHINE- READABLE STRUCTURES COST CONTROL AND OPTIMIZATION
  • 16.
    COMPLETELY NEW ROLE,NEVER EXISTED § Unique combination of skills § Need real experience § Hard to find on the market Prompt Engineering Role
  • 17.
    EXTREMELY IMPORTANT § Dataencryption at rest and in transit § Access control § Data anonymization § Compliance with privacy laws § Incident response plan § AI model transparency and auditability § User consent and data rights § Limit the use of customer data for model training Privacy and Security
  • 18.
    NEW TECHNOLOGIES COMEWITH RISKS § OpenAI platform availability and outages § Unpredicted model behavior changes § Unstable results based on input data § Degraded performance at peak hours § Robust testing protocols § Model validation and verification § Continuous monitoring and feedback loops § Version control and model management § Caching and failover § Performance profiling and optimization Quality Assurance, Stability, Performance
  • 19.
    Gen AI CostsCan be Significant § Transactional costs / per token § Cost profiling § Using different models (price differ x10, or x100) § Pre-processing with cheaper models § Custom models to run on-premise § Constant monitoring and improving Cost Optimization and Control
  • 20.
    HOW TO UNDERSTANDA ROBOT? § Ask for Structured Data Formats (JSON/XML, etc.) § Additional AI models (NER and others) for data extraction § NLP Libraries (e.g. NLTK) § Built in quality control GPT Responses vs. Machine-Readable Structures
  • 21.
    Welcome to the NewEra of GenAI
  • 22.
    VIRTUALLY EVERY SYSTEMWILL USE IT COMPANIES NOT USING GENERATIVE AI WILL BE LOSING IN COMPETITION COMPANIES WILL NEED TO GAIN GENAI EXPERTISE Future of Business with GenAI 1 2 3 LARGE VENDORS WILL OFFER NEW TOOLING 4
  • 23.
    § Plan forthe rise of competition § PoC can be done in weeks or even days § All tools are out there § # Years in AI business § Own AI-powered solutions § Expertise in AI, prompt engineering § Expertise in your domain STRATEGICALLY THINK ACT QUICKLY EVALUATE AND SELECT YOUR AI PARTNER Getting Your Business Ready for GenAI
  • 25.
    Subscribe to OurNewsletter on LinkedIn DON’T MISS THE NEXT EDITION!
  • 26.
    Thank You! +1 (847)559-0864 sales@velvetech.com www.velvetech.com