© Smartbots
introduction
There's been a lot of hype around Artificial Intelligence over the past 5 years. Starting around 2013, companies have
attempted to launch basic chatbots to lower costs and improve customer experiences. However, due to immature Natural
Language Processing algorithms and poor design in the early years, many have failed. Our discussion will explore how
conversational AI has matured in the last 5 years and demonstrate how a few key advances in NLP/NLU (Natural Language
Understanding) is closing the intuition gap.
Key takeaways:
- Key approaches to identify high-value use cases for chatbots in wholesale/distribution
- How conversational AI can fit into a larger AI strategic roadmap
- Live workshop to develop a few use cases
- How to develop KPIs to measure the effectiveness of chatbots
© Smartbots
Mike Echlin
VP of Customer Success
Smartbots.ai
AI Evangelist
Conversational AI focus
Yatin Dhareshwar
VP - Digital Services
YASH Technologies Inc.
Digital Transformation
Enabler
Connected Solutions focus
speakers
© Smartbots
Conversational AI 2020
Smartbots Labs - Research
© Smartbots
2000 2010 2020
TechAdvancement
Web Applications
Mobile Apps
Conversational AI
We are in the age of AI driven conversations
© Smartbots
what do EXPERTS SAY?
Conversational AI is next frontier
© Smartbots
what do EXPERTS SAY?
By 2022, 70% of white-collar workers
will interact with conversational agents
on a daily basis
Ref:
https://www.gartner.com/smarterwithgartner/chatbots-wil
l-appeal-to-modern-workers/
54% of US online consumers believe that
interacting with a chatbot will have a negative
impact on their quality of life
Ref:
https://go.forrester.com/blogs/forresters-latest-infographic-consum
ers-dont-believe-the-chatbot-hype/
Conversational AI is next frontier; have realistic expectations
© Smartbots
conversational AI POTENTIAL
Right execution gives BIG results
Customer Satisfaction
Operating Efficiency
Cost
Query Resolution Time
© Smartbots
© Smartbots
Breakthrough technology
© Smartbots
conversational AI EVOLUTION
1st Gen
2nd Gen
3rd Gen
4th Gen
5th Gen
Rule based
Fixed Flow
Dynamic
Flow
Open
Domain
Learn from
Humans
src: https://go.rul.ai/vaam
© Smartbots
2nd Gen EXAMPLE
Flow changes cannot be handled
© Smartbots
3rd Gen EXAMPLE
Can handle multiple flows
© Smartbots
conversational AI EVOLUTION
3rd Generation Bots are perfect fit for Enterprise
use-cases
© Smartbots
business Applications
Customer
Support
HRM CRMERP/SCM
© Smartbots
From this...
© Smartbots
“I want to request time off”
To this...
© Smartbots
calculating ROI
Improved customer satisfaction
Evalue CSAT score on chat/voice bot channel through survey
Monitor Loyalty
Watch for changes in Net Promoter Score with the new channel introduction
Reduction in Cost Per Query
Compare the cost per query between bots & human agents
Efficiency
Measure the time saved for the human agents
Come up with ROI calculator at the beginning
© Smartbots
conversational AI MUST HAVE
1. Ability to understand language
a. Domain knowledge
b. Enterprise knowledge
c. Small talk
2. Ability to perform action
a. Read/ write information from backend
b. Process Automation
3. Ability to engage user
a. Context aware - graceful fallback, sentiment analysis
b. Memory - training log data
c. Learning
AI is must for Conversational Agents
© Smartbots
Sample use cases SUPPLY CHAIN / DISTRIBUTION
Use case examples for supply chain, distribution and wholesale.
❖ JIT product delivery
Grocer is running a promo on certain products and wants to know stock levels and availability.
"how soon can I get 18 cases of x?"
❖ Simplifying software usage
Complex searches and navigation
"Show me products in this family"
Technical reference guides
"Which parts do I need for this HVAC unit replacement?"
© Smartbots
cont’d Use Cases
❖ Product recommendations
"I'm looking for a solution for my warehouse rack"
❖ Call center efficiency
Top 10 FAQs
❖ Internal process and solutions
On-boarding training bot
Policy FAQ
❖ Transportation
Driver updates - “ Due to road construction, the delivery will be 30 min late” sent as a text to logistics chatbot
© Smartbots
bot BUILDING PROCESS
Use-case
Identification Design Implementation Test & Train Deployment
© Smartbots
AI Operational Framework
Business Value AI Roadmap
Success MetricsPOCs
● Identify areas with high friction (ERP?)
● Build a master list of potential use cases
● Create KPIs and/or success factors
● Appoint AI product owner
● Incorporate all types of AI (speech, vision,
images, deep learning)
● Prioritize POCs and socialize AI strategy
● Establish 12-24 mo. budget
● Prioritize use cases that can be measured and
completed within one quarter
● Identify SMEs and negotiate buy-in
● Leverage high-value POCs to demonstrate
quick wins
● Establish bot metrics baselines
● Set expectations appropriately
© Smartbots
socializing AI in the Org
● 100% control of BOT knowledge and learning curve
● Creating efficiencies with repetitive tasks
● NOT building The Terminator
● Long-term AI strategy should demonstrate evolution and gradual maturity of chatbots
© Smartbots
discussion Points - Top 5 Questions CIO’s ask
● What are some of the biggest challenges and pitfalls?
● What your teams will need to succeed?
● How to define success? (18)
● What are the leading Tools, Platforms, Licensing?
● How to nurture organizational adoption? (23, 24)
© Smartbots
“To assess AI’s value and risks, CIOs need
to set realistic expectations for AI”
- Svetlana Sicular, Vice President Analyst, Gartner (Sept 2019)
© Smartbots
thank YOU
Future is here
www.smartbots.ai info@smartbots.ai +1-800-540-7723
© Smartbots
workshop toolkit (example)
Business
Process/Department
Friction Pain Points Dialog Notes
Customer Service Product A Low CSAT “I’m missing parts”
Marketing
Systems Friction Pain Points Dialog Notes
ERP
Channels Friction Pain Points Dialog Notes
Suppliers

Conversational AI - 2020

  • 1.
    © Smartbots introduction There's beena lot of hype around Artificial Intelligence over the past 5 years. Starting around 2013, companies have attempted to launch basic chatbots to lower costs and improve customer experiences. However, due to immature Natural Language Processing algorithms and poor design in the early years, many have failed. Our discussion will explore how conversational AI has matured in the last 5 years and demonstrate how a few key advances in NLP/NLU (Natural Language Understanding) is closing the intuition gap. Key takeaways: - Key approaches to identify high-value use cases for chatbots in wholesale/distribution - How conversational AI can fit into a larger AI strategic roadmap - Live workshop to develop a few use cases - How to develop KPIs to measure the effectiveness of chatbots
  • 2.
    © Smartbots Mike Echlin VPof Customer Success Smartbots.ai AI Evangelist Conversational AI focus Yatin Dhareshwar VP - Digital Services YASH Technologies Inc. Digital Transformation Enabler Connected Solutions focus speakers
  • 3.
    © Smartbots Conversational AI2020 Smartbots Labs - Research
  • 4.
    © Smartbots 2000 20102020 TechAdvancement Web Applications Mobile Apps Conversational AI We are in the age of AI driven conversations
  • 5.
    © Smartbots what doEXPERTS SAY? Conversational AI is next frontier
  • 6.
    © Smartbots what doEXPERTS SAY? By 2022, 70% of white-collar workers will interact with conversational agents on a daily basis Ref: https://www.gartner.com/smarterwithgartner/chatbots-wil l-appeal-to-modern-workers/ 54% of US online consumers believe that interacting with a chatbot will have a negative impact on their quality of life Ref: https://go.forrester.com/blogs/forresters-latest-infographic-consum ers-dont-believe-the-chatbot-hype/ Conversational AI is next frontier; have realistic expectations
  • 7.
    © Smartbots conversational AIPOTENTIAL Right execution gives BIG results Customer Satisfaction Operating Efficiency Cost Query Resolution Time
  • 8.
  • 9.
  • 10.
    © Smartbots conversational AIEVOLUTION 1st Gen 2nd Gen 3rd Gen 4th Gen 5th Gen Rule based Fixed Flow Dynamic Flow Open Domain Learn from Humans src: https://go.rul.ai/vaam
  • 11.
    © Smartbots 2nd GenEXAMPLE Flow changes cannot be handled
  • 12.
    © Smartbots 3rd GenEXAMPLE Can handle multiple flows
  • 13.
    © Smartbots conversational AIEVOLUTION 3rd Generation Bots are perfect fit for Enterprise use-cases
  • 14.
  • 15.
  • 16.
    © Smartbots “I wantto request time off” To this...
  • 17.
    © Smartbots calculating ROI Improvedcustomer satisfaction Evalue CSAT score on chat/voice bot channel through survey Monitor Loyalty Watch for changes in Net Promoter Score with the new channel introduction Reduction in Cost Per Query Compare the cost per query between bots & human agents Efficiency Measure the time saved for the human agents Come up with ROI calculator at the beginning
  • 18.
    © Smartbots conversational AIMUST HAVE 1. Ability to understand language a. Domain knowledge b. Enterprise knowledge c. Small talk 2. Ability to perform action a. Read/ write information from backend b. Process Automation 3. Ability to engage user a. Context aware - graceful fallback, sentiment analysis b. Memory - training log data c. Learning AI is must for Conversational Agents
  • 19.
    © Smartbots Sample usecases SUPPLY CHAIN / DISTRIBUTION Use case examples for supply chain, distribution and wholesale. ❖ JIT product delivery Grocer is running a promo on certain products and wants to know stock levels and availability. "how soon can I get 18 cases of x?" ❖ Simplifying software usage Complex searches and navigation "Show me products in this family" Technical reference guides "Which parts do I need for this HVAC unit replacement?"
  • 20.
    © Smartbots cont’d UseCases ❖ Product recommendations "I'm looking for a solution for my warehouse rack" ❖ Call center efficiency Top 10 FAQs ❖ Internal process and solutions On-boarding training bot Policy FAQ ❖ Transportation Driver updates - “ Due to road construction, the delivery will be 30 min late” sent as a text to logistics chatbot
  • 21.
    © Smartbots bot BUILDINGPROCESS Use-case Identification Design Implementation Test & Train Deployment
  • 22.
    © Smartbots AI OperationalFramework Business Value AI Roadmap Success MetricsPOCs ● Identify areas with high friction (ERP?) ● Build a master list of potential use cases ● Create KPIs and/or success factors ● Appoint AI product owner ● Incorporate all types of AI (speech, vision, images, deep learning) ● Prioritize POCs and socialize AI strategy ● Establish 12-24 mo. budget ● Prioritize use cases that can be measured and completed within one quarter ● Identify SMEs and negotiate buy-in ● Leverage high-value POCs to demonstrate quick wins ● Establish bot metrics baselines ● Set expectations appropriately
  • 23.
    © Smartbots socializing AIin the Org ● 100% control of BOT knowledge and learning curve ● Creating efficiencies with repetitive tasks ● NOT building The Terminator ● Long-term AI strategy should demonstrate evolution and gradual maturity of chatbots
  • 24.
    © Smartbots discussion Points- Top 5 Questions CIO’s ask ● What are some of the biggest challenges and pitfalls? ● What your teams will need to succeed? ● How to define success? (18) ● What are the leading Tools, Platforms, Licensing? ● How to nurture organizational adoption? (23, 24)
  • 25.
    © Smartbots “To assessAI’s value and risks, CIOs need to set realistic expectations for AI” - Svetlana Sicular, Vice President Analyst, Gartner (Sept 2019)
  • 26.
    © Smartbots thank YOU Futureis here www.smartbots.ai info@smartbots.ai +1-800-540-7723
  • 27.
    © Smartbots workshop toolkit(example) Business Process/Department Friction Pain Points Dialog Notes Customer Service Product A Low CSAT “I’m missing parts” Marketing Systems Friction Pain Points Dialog Notes ERP Channels Friction Pain Points Dialog Notes Suppliers