Artificial Intelligence in Business
An Architect’s Viewpoint
CWIN San Francisco
Michael Martin
December 7, 2017
Education
Michael C Martin
Enterprise Architect
BS Computer Science
Experience
• Capgemini
• IBM
• Apple
• GE
• Coca-Cola
• DCA / Attachmate
Fun Facts
• Outgrew Height Requirements for
Airforce ROTC Scholarship
• Started work for IBM at 19
• Been involved with IT since 1980
• Love solving puzzles and putting
processes and systems in place that
make a difference
Projects
• Ecommerce (QVC, Macys.com,
NBA.com)
• Content (DC Comics, Coca-Cola)
• Infrastructure (Michael Kors)
• Mobile (Quintiles, Cox
Communications, Kohls.com, Coca-
Cola)
Dimensions of Artificial Intelligence
10. Cognition
8. Deep
learning
9. Image
analysis
6. Knowledge engineering
7. Natural language
generation
5. Machine
learning
1. Robotics
2. Sensory
perception
3. Speech recognition
4. Natural
language
processing
Artificial
Intelligence
Interesting Intersections
Data:
• Prior:
• Siloed content
• Structured Content
• Linear processes
• Recent:
• Data warehouses
• Data “Lakes”
• More opportunities to collect content
• Now:
• Unstructured Content
• Streaming content
• Constantly moving content
• Opportunities for cross collaboration
Interesting Intersections
Infrastructure:
• Prior:
• Internal Systems
• Siloed Systems
• Lack of integration or alignment
• Recent:
• Integration busses / Web Services
• “Master Data Management”
• Now:
• Robust Cloud deployment tools
• On-demand infrastructures
• Federated environments
Data
Infrastructure
Interesting Intersections
Tools and Platforms:
• Prior:
• Intra-system reporting tools
• Summary reports
• Green bar paper
• Recent:
• Business Intelligence Systems
• Dashboards / KPI’s
• Structured Data tooling
• Departmental ad-hoc query tools
• Now:
• “executive ready” BI tools
• Unstructured data tools
• External Systems data aggregators
• API services and contracts
Data
Infrastructure Tools
Interesting Intersections
Data
Infrastructure Tools
AI
Tools and Platforms:
• Prior:
• Intra-system reporting tools
• Summary reports
• Green bar paper
• Recent:
• Business Intelligence Systems
• Dashboards / KPI’s
• Structured Data tooling
• Departmental ad-hoc query tools
• Now:
• “executive ready” BI tools
• Unstructured data tools
• External Systems data aggregators
• API services and contracts
AI and the Diffusion of Innovation Curve
Source: Wikipedia - Rogers Everett - Based on Rogers, E. (1962) Diffusion of innovations. Free Press, London, NY, USA. /
https://en.wikipedia.org/wiki/Diffusion_of_innovations
The Five Senses of Artificial Intelligence
Source: Capgemini, “The five senses of Artificial Intelligence: Christopher Stancombe”, May 2017
AI and Business Value
Effectiveness
Model
Innovation
Model
Efficiency
Model
Expert
Model
Data
Complexity
Work
Complexity
Automate
Augment
Examples where AI can assist
Legal Research Medical Research Fraud Detection
ComplianceCustomer SupportSocial Media
Systems Operation
Safety and
Surveillance
Expert Systems
Processes where data is looked at in real time over a stream of content,
structured data, or non-structured data.
How the AI projects start …
Source: http://dilbert.com/strip/2016-06-20
EDIT 2017 12 12
Image removed in consideration of licensing terms. Please refer to the URL
below for the referred to image
Thinking through the AI problem
What are the bodies of content that I can use to help
combine to come up with for analysis?
Content
Context
Sentiment
Intent
What is the domain that the information relates to?
What lens should the information be viewed? Are there
external influences to take into account?
What is the desired outcome / hypothesis that we can
validate by doing analysis?
AI Reference Architecture – Logical View
Big Data
Store
Data
Transformation
Analytical
Processing
Storage and
Integration
Production
Systems
Extract
Transactions
Campaign +
Inbound Xactions
Customer Product
Master
Unstructured
Data
Access Data
Transform
Learning
Scoring
Mining
Modeling / Inputs
Results Store
Integration
Queries /
Analytics
Post Processing
Systems of
Consumption
API Contracts
Web / Mobile /
Dashboards
CRM / Customer
Experience
Curation /
Modeling /
Queries
Business
Intelligence
Microsoft Provided Algorithm Cheat Sheet
Source : https://docs.microsoft.com/en-us/azure/machine-learning/studio/algorithm-cheat-sheet
Capgemini Smart Analytics™ Platform
Rapid prototyping of Cognitive and other AI use cases behind your firewall
To enable better
social customer
service
To proactively
provide information
of prospect to
financial advisors
To enable predictive
intelligence to
customer data
To provide
relevant next
best offers
To analyze voice
of customer in
near to real
time
To enable
location
based
offering
Connecting
all customer
touchpoints
https://www.capgemini.com/insights-data/smart-platform-delivering-insights-for-your-business
IBM Watson plugin for Wordpress – Sentiment
analysis
Source: https://thecustomizewindows.com/2017/06/wordpress-plugin-to-analyze-post-emotion-with-ai-ibm-watson/
Macy's tests artificial intelligence as a way to
improve sales
Source : http://www.latimes.com/business/la-fi-agenda-macys-20160725-snap-story.html
Interesting Directions and Non-Functional
Perspectives
Distributed Machine
Learning
Privacy, Ownership,
Nexus of Data
Federation, Licensing,
Partnerships
Curation, Ownership
Increasing Public Awareness
à Regulation?
Lifecycle Management,
Financial Commitment
Do not underestimate
resources/time to manage (find,
source, clean, govern) datasets for
AI projects
AI projects will require a culture change,
and sometimes real surprises. Driving the
corresponding transformation will require
high level of sponsorship and
willpower to drive change.
Communicate and educate
at each step taken during the
project.
You need to build trust with
the machine!
The AI era is now!
AI is a disruption factor. It is also
a learning curve. The sooner you
get on it, the better you can
manage your own destiny as a
business.
A successful AI project can
only be built with an
integrated Business/IT
team. The best use cases are
problems that sometimes
exist since forever, but never
found a real answer.
… but data & AI techniques are
only means to an end, not the end
itself!
Think Big, Start Small : 80 % of our AI projects show relevant
and actionable results in 3 months
Closing thoughts from the Architect
PRO:
• Potential to realize underused data,
provide a competitive edge in your
market
• Opportunities for those who embrace
and help derive business value from
new processes
CON:
• High probability that there will need to be
changes in your processes and systems to
accommodate new information flows and
analysis opportunities.
• Some jobs may be affected more than
others
Watch:
Learn:
Unless your business is in the business of AI / ML (startup, moonshot program), hang
back a little to see how the tools and processes will mature over the next few years.
However, you should keep a close eye on this, possibly run some pilot projects to
have a short term goal in mind and to see what opportunities there are to integrate
this into your systems — this is not a question of if — it’s a question of when this will
be commonplace and table stakes.
Capgemini: AI – The successful Implementers Toolkit
https://www.capgemini.com/consulting/resources/ai-the-successful-implementers-
toolkit/
IBM Watson: Everything You Ever Wanted to Know
https://www.fool.com/investing/2017/08/30/ibm-watson-everything-you-ever-wanted-
to-know.aspx
Resources