88% of AI projects end up not making it into production. Discover how Google Dialogflow, built by Google Cloud AI, can help you inject Conversational AI into your workstreams to delight your customers and clients. Working with Google solutions can help address real world problems in various industries so AI initiatives do not end up sitting on a shelf.

1. Tony Lusk
Customer Engineer,
Google Cloud
Conversational AI
Moving off of your roadmap
and into your workstreams

2. Confidential & Proprietary
Background

3. Google is an AI company

4. Gmail Smart
Reply
12%
of all
responses
sent on
mobile

6. 21M
Developers
1000's
Deep learning
researchers
~1M
Data scientists
Who can
actually use
AI today?
Very few users today can
create a custom ML model
We need to make AI
accessible to millions more

7. How
Google
can inject
AI into
your
business
Building blocks Platform Solutions

8. A comprehensive set
Sight Language Conversation
Cloud Speech-to-Text
Dialogflow Enterprise Edition
Cloud Text-to-Speech
Cloud Vision
Cloud Video Intelligence
Cloud Translation
Cloud Natural Language
New
AutoML
Natural Language
NewAutoML TranslationAutoML Vision New
of AI building blocks

9. What is a “conversational
experience”?

10. Proprietary & Confidential 10
Demo
Trip to
Mexico

11. Conversation is the new UI!
80%
of customer
interactions can be
resolved by
well-designed bots
– Accenture
60%
of consumers want
easier access to
self-serve solutions
for customer service
– Ovum
50%
of enterprises will
spend more on bots
than traditional mobile
app dev by 2021
– Gartner

12. Consumers have high expectations
Based on data collected in Fall 2017 from sample of 1,051 U.S. adults, ages 18-64.

13. Meeting
expectations
is hard
without the
right tools
NLU doesn’t
grow on trees!
Few companies have
the expertise to do
NLU as well as
consumers expect
Maximizing reach
takes effort
Building support for
multiple languages,
platforms, devices, and
apps is complex
Enterprise
integration is critical
Integration with backend
services (eg CRM)
requires open, flexible
infrastructure

14. What is
Dialogflow
Enterprise
Edition?
● The emerging standard for building natural and rich
conversational experiences for customer service and
other applications across devices, websites, messaging
platforms, and apps
● Powered by Google AI
● Built on Google Cloud Platform infrastructure, with Cloud
Support and SLA available
● Multilingual and platform-agnostic: 20+ languages
supported, 14 single-click integrations, and 7 SDKs

15. Dialogflow benefits for users
Build faster Engage more efficiently Maximize reach
Start training with only a few
examples
40+ pre-built agents and
“small talk” features
Go-to-market with a simple
bots in hours
Built-in, world-class natural
language understanding
Multiple fulfillment options,
including built-in code editor
Training and analytics
across platforms
Build once, deploy
everywhere
20+ languages supported
(most in product category)
14 single-click platform
integrations & 7 SDKs

16. Conversation management
Text
Digital Voice
Phone calls through gateway
Dialogflow
Enterprise Edition
Google Cloud Orchestration
Directory
CRM
Other systems
& APIs
Platform
AppEngine
Infrastructure
Compute Engine
Microservice
Cloud Function
Multi Channel
Fulfillment
Phone
Mobile app
Car
TV
Google home
Chat on web
Facebook Messenger
Twitter
Google Assistant on mobile
Completing the picture

17. Confidential & Proprietary
Customers

18. Dialogflow customers across industries

19. ● Serves 30+ million passengers annually, operates 200+
aircraft, and flies to 163 destinations worldwide
● Its “BB” customer service virtual agent lets customers
find/book flights via FB Messenger, with handoff to live
agents for complex requests
● Key quote: “Dialogflow was very easy for our developers
to use. We were able to build and ship the experience
really quickly."
Example Use Case :
Industry: Travel

21. Proprietary & Confidential 21

22. Confidential & Proprietary
Momentum

23. Developer adoption
is growing quickly
No.ofDevelopers
2016 Cloud Next
‘17 (March)
Cloud Next
‘18 (July)
600,000+
150,000+

24. Confidential & Proprietary
What’s New
Contact Center AI

25. 7 out of 10 U.S.
consumers say
they’ve spent more
money to do business
with a company that
delivers great
service.*
* American Express 2017 Global
Customer Service Barometer
Give customers access to answers to common
questions 24x7 globally, without queuing
Help live agents answer questions more quickly
and accurately, with fewer/shorter customer holds
Make complex, frustrating IVR trees obsolete
without having to rely more on live agents
Top 3 ways that AI can help contact centers
provide great customer service:
1
2
3

26. Use cases in the contact center and beyond
● Customer service
● E-commerce
● Home entertainment
● Auto
● Organizational knowledge
● Onboarding & Helpdesk

27. Contact
Center
Provider
Contact
Center
Interface
Virtual Agent
Agent Assist
Knowledge
Base
(PDF/HTML)
Phone
Chat
Backend
Fulfillment
Agent
Desktop
Agent
Customer
Reference architecture

28. New features relieve contact center pain
Phone Gateway Beta Knowledge Connectors Beta
Built-in sentiment analysis Beta
Built-in Text-to-Speech Beta
Automatic Spelling Correction Beta
Lets you instantly convert any Dialogflow
virtual agent into a conversational IVR
Uses technology similar to that used in
Google Search to correct for user typos
Complements explicit intents with automated
responses sourced from knowledge bases
Measures sentiment to provide a data point
about need for live agent handover
Powered by DeepMind’s WaveNet to deliver
high-quality speech responses for telephony & IoT

29. Phone Gateway
Lets you instantly convert any
Dialogflow virtual agent into a
conversational IVR (Interactive Voice
Response) system.
Based on the same Google
infrastructure used by Google Voice,
Project Fi, Google Hangouts.
Telephony service, speech
recognition, NLU, and speech
synthesis are handled
out-of-the-box.

30. Knowledge
Connectors
Dialogflow lets you pre-build explicit
responses to specific user intents.
Knowledge Connectors enrich the
conversation by complementing explicit
intents with automated responses
sourced from internal knowledge bases
(articles, FAQs, etc.).

31. Confidential & Proprietary
Automatic Spelling
Correction
Users commonly use abbreviations
and incorrect spelling or grammar,
particularly when typing in a hurry.
Our new Automatic Spelling Correction
feature uses technology developed for
Google Search to correct for user
typos.

32. Built-in
Sentiment Analysis
Measures sentiment of a user request
to provide a data point about the user’s
emotional state of mind.
That data can be used to help determine
the need for bringing in live agent help.

33. Built-in
Text-to-Speech
Dialogflow uses Cloud
Text-to-Speech powered by
DeepMind’s WaveNet technology
(closes audio-quality gap with human
speech by 70%) to deliver natural,
precise speech responses.
Useful for telephony as well as IoT
applications (e.g., cars, TVs).

34. That’s a wrap.
Questions?