Nella
ChatBOT
for Customer Service
Businesses are under relentless pressure to
cut down cost, and to do so, will adopt
technologies that are inspired by artificial
intelligence (AI).
Value Proposition
BOT platform will fundamentally change the way
we interact with computers, people and
enterprise scale systems. BOTSolutions will be
based on advanced textand content analytics,
machine learning, natural language processing
and intelligent knowledge processing engine.
Conversational interfaces (BOTsolution) built using
BOTplatform to providetransformational
benefits resulting in:
Next generation machine
interaction to improve
customer experience -an
ability toprovide reliable,
cost-effective and responsive
services.
Self-Service (“Do-It-Yourself”)
option to help reduce agent
interaction through phone
calls, emails and personal
interactions.
Increase the overall
productivity and reduce
cost of operations.
Features
CanIntegrate seamlessly
with existing customer
support ITsystem
(Customer CarePortal)
Feature rich speech
recognition, AI and natural
language processing
capabilities
Canbe easily adapted to variety of
use cases acrossenterprises
- For resolving ITapplication support
related issues (replaces L1 Desk)
- Self Service assistant for Clinical
Laboratory (Appointment booking,
Queries on Test procedures etc.)
- Personal Banking FAQs
Knowledgebase (Ontology)
built through historical
data can be easily
extended
Flexible commercial model
with cloud / On-Premise
deployment options.
Solution Architecture
User Utterance
Bot Solution
Business
Requirement &
raw /organized
business data
Failure Analysis
Knowledge
base/FAQ
Administration
Knowledge Base
Build Bot Solution – FAQ & Dialog
definition for tasks to be performed
in consensus & discussion with
Business (Manual Approach)
Business needs
customer service
automation
User
Live Agent
Automation to build
the Bot Solution
Automated
User Profiling
Bot performance
Statistics
Admin Console & Controlled Training
Training Actions
based on Failure
Analysis (Manual)
Live chat
Seek
information
or perform
some task
Live Agent Dashboard
Handover to
Live Agent
Bot Training
(Manual)
Machine learning for User
profiling and suggestions.
Sentiment
Analysis
Context
Detection
Resolution
/ Response
Architecture
User
Utterance
UserQuestion
NLUEngine
Above
Threshold
BusinessRules
Question
Expansion
Natural
Language
Processing
Conversation
Engine
Clarification
question
Word Net
Knowledge Base Below
Threshold
DialogContext
Extraction
Clarification
Question
UI
Experience
Conversation
Engine
Dialogue
Definition
Knowledge
Extraction
Backend
Connectors
Integration Architecture
Internet
HTTP/
HTTPS
Conversation
Interface
ServiceGateway
REST
APIs
JPA /JMS
JDBCetc.
SOAP/
REST
ITSM
(Remedy, HP
Service
Manageretc.)
CRM,ERP,SCM
systems
External
Systems (Wiki,
Weather etc.)
Cloud/ On-PremiseDeployment
Prebuilt connectors–
ITSystem
BOT Client
Key Features & Roadmap
Selfand
Controlledlearning
• Self learns from previous
conversation to guide user
with appropriate utterance.
• Failures Analysis through
administrative panel for
identifying & training the
BOT to deal with new
user intents.
• Understands User profile
based on historic conversations
& converses based on the
user preferences.
• Multi channel Integration with
Knowledge base using RPA –
Through Admin console/ Through
files repository / Web Page
Repository
• Integration with Orchestration/
RPA Products for Service
Automation
• Integration with Ticketing Tools
e.g. Service Now, Fresh Service,
Zendesk
• Consume AI/ML models in real
time as micro services to enable
Verizon decision making systems
(e.g. PEGA, Sales force, AEM) to
provide insights driven decisions
for systems of engagement
Integration
Capability
• Mobile App
• Integration with
Messengers e.g.
Skype, FB Messenger
• Integration with IVR
• Smart device /Sensor
integrations – Alexa ,
Google home
UserInterface
Channels
• Ambiguity Resolution
• Ask Question differently
• Provide recommendations
• Guided conversation
• Conversation through
Interactive UI Cards
• Image Learning
Conversational
AI
• Multilingual Support
• Text as well asvoice
conversation
• Provide Information in Text,
Media &Document output
format.
• Verification usingOTP
• Handover to live agent –
On explicit user demand/
Live Agent Monitoring the
BOT Conversations/ Based on
Sentiment Analysis of the
progression of conversation.
Core
Features
Business Understanding &
Problem Statement (Use
Cases)
Collect Knowledge Base &
Intent Data
 Kinesis
 Kafka
Data Preparation and
Cleaning
 Cassandra/Neo4j
 HDFS/Amazon S3
 Elastic Search/Oath
Data Visualization and
Analysis
 Apache Spark
 Flink
 EMR
Feature Engineering
Knowledge Base Creation
Model Training and Prediction/
Recommendations
 H2O Driverless AI/Sage
Maker
 Scikit-learn/Spark Mllib
 TensorFlow/Jupyter N/B
Model Evaluation
Model Deployment
 Flask API
 Docker Container
 Kubernetes
 Ngnix
 H20 MoJo Object
 Apache Thrift
 Git
Monitoring and
Debugging
 Tableau
 Kibana
 Amazon Quicksight
Use Case
exit criteria
Controlled Training
Feature Augmentation
Data
Augmentation YESNO
Data Sources
Documents
Historic
Conversations
Web Pages
Through Admin
Console
Automated Feature Engineering
Self Learning
Perform task/Seek
Information
AI / ML Factory
What’s in it For
Admin Console to monitor Nella performing,
administer FAQs
Handover to live agent
Track the progress on the reported Issue/Incident
Update Incident details
Quick Query about the incidents
Notifications about missing Acknowledgement & Resolution SLAs
Auto Assignment of Incidents
Track tickets & Quick query
Admin Console
ITIL Support
Auto creation of Problem Record based on recurring incidents
Critical/High priority Incident Management
Notifications about missing Incident updates
End user Customer service team
Search information from Incident History,
knowledge base documents &FAQs
PasswordResets
Validation usingOTP
SoftwareInstallation
Report an Issue/Incident
Update Incident details
Know about progress on the reportedIssue/Incident
Ask Question/ SeekInformation
Request Reports/Documents
Ticket -Reporting & Enquiry
Self Help
Ask Question
Converse & get the task done/seek information.
NO WAITING
i i
Key Differentiator
Strong NLP
Capabilities
Easyintegration
with backend
AI/MLDriven
Intelligence
Complete On-premise
Deployment
The platform is light weight,scalable, secure and can be easily deployed on on-premise infrastructure. The platform
Adheres to GDPR compliance
Platformis driven by strong machine learning algorithms running behind the scene. Bot has in-built knowledge
extraction module to answer open domain questions as wellas FAQs.
• The platform has been specifically designed to meet enterprise needs as against generic which clearlylacks supporting
complex conversations. For example-
• User switching the context allof sudden (most ofthe platform willforget the original context and hence looses the track.)
• Retain context by easy configuration. (itcan store specific information provided by user in his/her earlierconversation)
• Create domain dictionary(synonyms,custom entities etc.)
• Easy handovertohuman agent in case user is not comfortablein carrying over the conversation with bot.
• The configuration allows user toeasily integrate with backend system and also post process response data using highly
customisable groovylanguage.
What is Chat Bot ?
Answer
Connect users to right
resource, person and
service.
Communicate
A umanized, effortless
experience, natural way of
expressing and
communication
Guide
Monitor every interaction
and tune the virtual
assistant accordingly for
continuous performance
improvement.
Listen
Listen and analyze the
user issues and provide
quick response.
Measure
Learns about users interests,
understand their behavior so
that a bot can respond to the
issues in most relevant way.
Engage
Engage 3 times more
than a traditional FAQ
or web-self-service.
Action (A) Bots
• Prebuilt bots for
ERP, CRM functions
• Functional Bots
• Service
• HR Bots
• Sales Bots
• Finance Bots
• Procurements
Bots
• Extend search with
knowledge
discovery tools like
Sinequa
• Intent discovery
using Solr
Knowledge (K) Bots
•Natural Language
Processing (NLP)
•Natural Language
Understanding (NLU)
•Natural Language
Generation
•Machine Learning
•Knowledge Engineering
(Ontology) Thesauri)
•Voice Bots
Text (T) Bots
NELL INFOTECH CHATBOT KIT
DIFFERENT CATEGORIES OF CHATBOT
MobileandPersonal
Assistants
Cognitive
Technologies
Conversational
Technologies
- Speech (Text to speech, Automatic Speech Recognition)
- Text-based chatbots
-Natural Language Processing, Machine Learning, AI
- Knowledge Management with Semantic Search
- Likes of Siri, Google Assistant, Facebook M, Cortana
- Support a horizontal range of individual tasks
Thank you!
Pleasegiveyourfeedback–
sheetal.jadhav@nellinfotech.com
connect@nellinfotech.com
Call: 9850088916/ 9890575963
Visitus – www.nellinfotech.com

Digital Transformation Services and Solutions - Chatbot Development

  • 1.
  • 2.
    Businesses are underrelentless pressure to cut down cost, and to do so, will adopt technologies that are inspired by artificial intelligence (AI). Value Proposition BOT platform will fundamentally change the way we interact with computers, people and enterprise scale systems. BOTSolutions will be based on advanced textand content analytics, machine learning, natural language processing and intelligent knowledge processing engine.
  • 3.
    Conversational interfaces (BOTsolution)built using BOTplatform to providetransformational benefits resulting in: Next generation machine interaction to improve customer experience -an ability toprovide reliable, cost-effective and responsive services. Self-Service (“Do-It-Yourself”) option to help reduce agent interaction through phone calls, emails and personal interactions. Increase the overall productivity and reduce cost of operations.
  • 4.
    Features CanIntegrate seamlessly with existingcustomer support ITsystem (Customer CarePortal) Feature rich speech recognition, AI and natural language processing capabilities Canbe easily adapted to variety of use cases acrossenterprises - For resolving ITapplication support related issues (replaces L1 Desk) - Self Service assistant for Clinical Laboratory (Appointment booking, Queries on Test procedures etc.) - Personal Banking FAQs Knowledgebase (Ontology) built through historical data can be easily extended Flexible commercial model with cloud / On-Premise deployment options.
  • 5.
    Solution Architecture User Utterance BotSolution Business Requirement & raw /organized business data Failure Analysis Knowledge base/FAQ Administration Knowledge Base Build Bot Solution – FAQ & Dialog definition for tasks to be performed in consensus & discussion with Business (Manual Approach) Business needs customer service automation User Live Agent Automation to build the Bot Solution Automated User Profiling Bot performance Statistics Admin Console & Controlled Training Training Actions based on Failure Analysis (Manual) Live chat Seek information or perform some task Live Agent Dashboard Handover to Live Agent Bot Training (Manual) Machine learning for User profiling and suggestions. Sentiment Analysis
  • 6.
  • 7.
  • 8.
    Key Features &Roadmap Selfand Controlledlearning • Self learns from previous conversation to guide user with appropriate utterance. • Failures Analysis through administrative panel for identifying & training the BOT to deal with new user intents. • Understands User profile based on historic conversations & converses based on the user preferences. • Multi channel Integration with Knowledge base using RPA – Through Admin console/ Through files repository / Web Page Repository • Integration with Orchestration/ RPA Products for Service Automation • Integration with Ticketing Tools e.g. Service Now, Fresh Service, Zendesk • Consume AI/ML models in real time as micro services to enable Verizon decision making systems (e.g. PEGA, Sales force, AEM) to provide insights driven decisions for systems of engagement Integration Capability • Mobile App • Integration with Messengers e.g. Skype, FB Messenger • Integration with IVR • Smart device /Sensor integrations – Alexa , Google home UserInterface Channels • Ambiguity Resolution • Ask Question differently • Provide recommendations • Guided conversation • Conversation through Interactive UI Cards • Image Learning Conversational AI • Multilingual Support • Text as well asvoice conversation • Provide Information in Text, Media &Document output format. • Verification usingOTP • Handover to live agent – On explicit user demand/ Live Agent Monitoring the BOT Conversations/ Based on Sentiment Analysis of the progression of conversation. Core Features
  • 9.
    Business Understanding & ProblemStatement (Use Cases) Collect Knowledge Base & Intent Data  Kinesis  Kafka Data Preparation and Cleaning  Cassandra/Neo4j  HDFS/Amazon S3  Elastic Search/Oath Data Visualization and Analysis  Apache Spark  Flink  EMR Feature Engineering Knowledge Base Creation Model Training and Prediction/ Recommendations  H2O Driverless AI/Sage Maker  Scikit-learn/Spark Mllib  TensorFlow/Jupyter N/B Model Evaluation Model Deployment  Flask API  Docker Container  Kubernetes  Ngnix  H20 MoJo Object  Apache Thrift  Git Monitoring and Debugging  Tableau  Kibana  Amazon Quicksight Use Case exit criteria Controlled Training Feature Augmentation Data Augmentation YESNO Data Sources Documents Historic Conversations Web Pages Through Admin Console Automated Feature Engineering Self Learning Perform task/Seek Information AI / ML Factory
  • 10.
    What’s in itFor Admin Console to monitor Nella performing, administer FAQs Handover to live agent Track the progress on the reported Issue/Incident Update Incident details Quick Query about the incidents Notifications about missing Acknowledgement & Resolution SLAs Auto Assignment of Incidents Track tickets & Quick query Admin Console ITIL Support Auto creation of Problem Record based on recurring incidents Critical/High priority Incident Management Notifications about missing Incident updates End user Customer service team Search information from Incident History, knowledge base documents &FAQs PasswordResets Validation usingOTP SoftwareInstallation Report an Issue/Incident Update Incident details Know about progress on the reportedIssue/Incident Ask Question/ SeekInformation Request Reports/Documents Ticket -Reporting & Enquiry Self Help Ask Question Converse & get the task done/seek information. NO WAITING i i
  • 11.
    Key Differentiator Strong NLP Capabilities Easyintegration withbackend AI/MLDriven Intelligence Complete On-premise Deployment The platform is light weight,scalable, secure and can be easily deployed on on-premise infrastructure. The platform Adheres to GDPR compliance Platformis driven by strong machine learning algorithms running behind the scene. Bot has in-built knowledge extraction module to answer open domain questions as wellas FAQs. • The platform has been specifically designed to meet enterprise needs as against generic which clearlylacks supporting complex conversations. For example- • User switching the context allof sudden (most ofthe platform willforget the original context and hence looses the track.) • Retain context by easy configuration. (itcan store specific information provided by user in his/her earlierconversation) • Create domain dictionary(synonyms,custom entities etc.) • Easy handovertohuman agent in case user is not comfortablein carrying over the conversation with bot. • The configuration allows user toeasily integrate with backend system and also post process response data using highly customisable groovylanguage.
  • 12.
    What is ChatBot ? Answer Connect users to right resource, person and service. Communicate A umanized, effortless experience, natural way of expressing and communication Guide Monitor every interaction and tune the virtual assistant accordingly for continuous performance improvement. Listen Listen and analyze the user issues and provide quick response. Measure Learns about users interests, understand their behavior so that a bot can respond to the issues in most relevant way. Engage Engage 3 times more than a traditional FAQ or web-self-service.
  • 13.
    Action (A) Bots •Prebuilt bots for ERP, CRM functions • Functional Bots • Service • HR Bots • Sales Bots • Finance Bots • Procurements Bots • Extend search with knowledge discovery tools like Sinequa • Intent discovery using Solr Knowledge (K) Bots •Natural Language Processing (NLP) •Natural Language Understanding (NLU) •Natural Language Generation •Machine Learning •Knowledge Engineering (Ontology) Thesauri) •Voice Bots Text (T) Bots NELL INFOTECH CHATBOT KIT
  • 14.
    DIFFERENT CATEGORIES OFCHATBOT MobileandPersonal Assistants Cognitive Technologies Conversational Technologies - Speech (Text to speech, Automatic Speech Recognition) - Text-based chatbots -Natural Language Processing, Machine Learning, AI - Knowledge Management with Semantic Search - Likes of Siri, Google Assistant, Facebook M, Cortana - Support a horizontal range of individual tasks
  • 15.