Gabi Zodik

How A.I. assistants will transform your business

Director, CTO and Head of Development for Watson Assistant

IBM Research - Haifa

© 2018 IBM Corporation

Introducing

Watson

Assistant

SolutionsWatson Assistant Solutions

are conversational assistants

that help businesses transform

their customer experiences

through an in depth set of

capabilities that deliver

proactive and personalized

help that is portable to any

device while ensuring our

customers data is ALWAYS

private.

YOURAssistant

Your BrandExceptionally, tailored experiences

Your DataPrivate, personalized, secure

Your

CustomersConstant improvement with insights

“Developing a bot without measuring the behavior of your users is like shooting in the dark”

• Measure your impact/reach

• Know your users

• Learn and improve the design of your conversation flow

• Measure your success

Set of metrics include: Dashboard, Funnel, Engagement, Conversation, Success

Bot Analytics

Cognitive Profile - Characteristics

Definition and modeling of various world domain

Reasoning from semantic events into meaningful signals

Using machine learning for the recommendation engine

Simplified developer experience for skill developers

”Cold Start” for new users

E2E Watson Assistant personalized conversation based on cognitive

profile capabilities

Cognitive Profile

Modeled the ontology for the restaurant domain

Implemented a simple rule-based recommendation engine

Provided justification for the reasoning

Cognitive Profile - PoC

IFTTT Example

Watson / Presentation Title / Date 7

Turn on my outdoor light upon voice command using my WA in my car

Create custom voice

commands for yours

Watson Assistance.

available on Android Auto

This trigger fires

when you

say your phrase

If this then that

WA

IFTTT

Skill

Dan: “Please turn on the lights”

WA: “Ok. The light is on”

WA Creates IFTTT Trigger

Start interaction about news (e.g., top 5 sports) once I leave work. Offer to read top 5 Tweets/ podcast when

I’m in a traffic jam. Suggest a nearby gas station when the fuel level is at one quarter

Sport News

Sport News

Near gas stations

Top Twits

Select your skillThis action will select a

skill

If You exit an Area

Then Listen to Sport

News

If You exit an Area

Then Listen to Sport

News

If this then that

WA

News

Skill

WA: “ Hi Dan. Would you like to

hear the top sports news”

Dan: “Yes”

WA: “Here they are …………………

…………………………………”Dan leaves work

WA Consumes IFTTT Action

Enables

• Conversation generation/driving

• Reasoning (e.g., rules)

WA Knowledge BrokerWASubscription and

Agent Broker

IBM

Message

Hub

Janus

Graph

Watson Assistant Blackboard

IBM Cloud

Services

K&R - A Common Knowledge Store Tied to a Pub/Sub Mechanism (Alpha)

Edge capabilities are required by many

customers to enable device control and

interaction in disconnected mode

– Primarily driven by our automotive

customers

Provide hybrid architecture that

enables (Q2/2018 target):

– Graceful degradation in disconnected

scenarios – providing natural language

driven disconnected interaction for

core scenarios

– Seamless handoff between

disconnected and connected modes

– A unified development experience to

create skills that work both on the

cloud and on the edge

Auto

Media, sensors Custom container

Bus connector

Watson Assistant

Skills and Agents

Hybrid Decision Module

Edge x Server Synchronization

Watson Assistant

Watson IoT

at the Edge

MQTT

message

bus

Edge functions

Watson IoT for

AutomotiveWatson Assistant

WatsonSTT

Engine

ClassificationEngine

WatsonTTS

Engine

EntityExtraction

Engine

Bridge API

Component

IBM Watson Assistant is Based on a Hybrid Architecture(Edge + Cloud Architecture)

Generic Ontology Definition, Management and Reasoning

(KITT+EBA)

Profile

Ontology

+

Triggers

Context

Ontology

+

Triggers

Automotive

Ontology +

Triggers

Common Event Generating Knowledge Store

Retail

Ontology +

Triggers

Smart

Buildings/

Hospitality

Ontology +

Triggers

…

Generic Ontologies

Domain Specific

Ontologies

Ontology Based Conversational Skill Generation

Ontology Framework Components

Consumer Experience – Delight your customers with an

engaging and personalized context aware experience.

Customer Insights & Data – The data and insights you

gather about your customers through their interaction with your

personal assistant are yours.

AI Applications – Proactive apps that leverage the power of

IBM Watson to understand, reason, learn, and recommend.

Library of Skills – Developers contribute a growing library of

personal assistant expertise you can use to build cognitive apps

for your customers (e.g., weather, travel, IFTTT, navigation and

other third-party integrations).

Custom Skills – Deliver custom cognitive services ( e.g.,

service and support, targeted promotions, payment, reminders,

automation, client preferences).

Edge – Hybrid mode, works off-line

Conversation Analytics – Enables you to continuously

improve your offerings

To recap .. Benefits of IBM Watson Assistant

• How can enterprise preserve their brand in a

conversational world ? (trademarks, personalities, etc.)

• Extreme personalization (emotion, mood, demographics,

multi party conversations, etc.)

• Multi modal interactions (e.g., gesture, touch, etc.)

• Zero coding for developing new skills, learning from

learning from examples, patterns/aspects

• Bot to bot interactions in the future

• Federation of assistants, working across assistants

• Data privacy and security, overcoming human mistakes

• Moral issue

• Elder companion, what if the elder is not feeling well ?

• Life protection, someone mentioning a desire to

terminate his life

What does the Future hold ?

Modalities

Fle

xib

leS

tru

ctu

red

Bot2Bot

Protocol

Agents

Basic Copyright Notice & Disclaimer

©2018 This presentation is copyright protected. All rights reserved. You may download or print out a hard copy for your private or internal use. You are not permitted to create any modifications or derivatives of this presentation without the prior written permission of the copyright owner.

This presentation is for information purposes only and contains non-binding indications. Any opinions or views expressed are of the author and do not necessarily represent those of Swiss Re. Swiss Re makes no warranties or representations as to the accuracy, comprehensiveness, timeliness or suitability of this presentation for a particular purpose. Anyone shall at its own risk interpret and employ this presentation without relying on it in isolation. In no event will Swiss Re be liable for any loss or damages of any kind, including any direct, indirect or consequential damages, arising out of or in connection with the use of this presentation.