Information Classification :Public

Inventing and (or)

Commercializing AI
By Group 3
Manoj Mishra
Case Study Prashant Sehrawat
Nischay
AIAM Mandeep Singh
Under Guidance of :
Prof Prageet Aeron
Information Classification :Public

ABOUT NUMENTA VISION

We believe that
Leading the New
neocortical principles
Era of CPU-Based AI
will drive the creation
of machine intelligence

HISTORYthe Redwood Neuroscience

In 2005, three years after founding
Institute, Jeff Hawkins co-founded Numenta, along with Donna
Dubinsky and Dileep George. Building upon two decades of
neuroscience research and a growing collection of peer-reviewed
papers, Numenta has turned its focus to commercializing its brain-
based AI technology.
Information Classification :Public

Numenta

 $36 Million in funding by 2016

 No of Employees : 16
Dana Dubinsky Jeff Hawkins
 Slow user adoption
(Cofounder CEO) (Cofounder )

Leading the New Era of CPU-Based AI

Dilemma :
 Should Numenta stay the course and continue its focus on Science and
Technology development
 Should company revamp its fundraising and invest in “killer app”
 How aggressively they should defend their IP and market Numenta to
outside world
Information Classification :Public

FACTS
 Entrepreneur and Innovator.
About Jeff  Created and commercialized the Palm pilot.
Hawkins  Advanced handheld computing technology.
 Founder of Handspring and Numenta.

Hawkins' Theories on Machine Intelligence

 Hawkins's aim was to understand how the human neocortex worked and then create software
and build machines that worked on the same principle.
 In 2002, Hawkins founded the Redwood Neuroscience Institute in Menlo Park, CA (now part of
the University of California, Berkeley)
 In 2004, he coauthored the book, In his book, Hawkins depicted his Memory-Prediction
Framework of how the brain works.
 In March 2005, Hawkins, Dubinsky, and Dileep George, then a graduate student at Stanford
University and research fellow at the Redwood Neuroscience Institute, founded Numenta.
 The company aimed first to understand the principles that allowed the brain to create
intelligent behaviour and then eventually to create algorithms that implemented the theory.
They called both the theory and the technology Hierarchical Temporal Memory (HTM).
Information Classification :Public

FACTS
Hawkins believed that the main problem with traditional ANN approaches
Hawkins
was the divergence from our understanding of the brain, especially around
Belief
areas of data representation and the usage of the time dimension.

Numenta and Its Business Models

 2005—2008: Phase I—Broad Platform (First-Generation Algorithms)

 2008—2011 : Phase Il—Application-Focused Platform
 2012—2013: Phase Ill—Enterprise Sales
 2014—2015: Phase IV—Product and Showcase Applications
 2015—2016: Phase V—Tech Transfer Model Other Business Model Threads - i.e
Open source IP-Only Commercial Licensing In March 2016 Numenta offered six
different license types (Details are in exhibit 4)
Information Classification :Public

Numenta Licenses
6 Type of Licenses
Open Source
1 License
AGPL v3

2 Trial License Non open source license for experimentation – No commercial right

3 Start up License Commercial License target to HTM focused startup company

4 Royalty License Commercial License option for established companies

Custom License
3 for Software / IP
Commercial License if above are not good fits

Custom License
4 for IP Only
Commercial License to use our IP form your own HTM implementation
Information Classification :Public

Proposed Acceleration Strategy

Information Classification :Public

Numenta’s Neural Network classification

Information Classification :Public

Numenta Growth Options

Aggressive
1 approach Ahmad’s View : Scale the company

Raise company
2 Profile
Dubisky and Colligan view : by protecting IP

3 Stay course Hawkins : Stay the course and continue develop Science and Technology
Information Classification :Public

Proposal
Being the type of start up : It is needed the receive
cash flow
Governments :
Type of One time Sale
Organisation Big Organisation :
Annual / Recurring

Startup / Individaul:
Pay per use

Cost Structure
 Numenta had been quite good in innovation however there were issues around commercialization of HTM Technology. The technology can be
also commercialized by associating with critical aspects of cyber security where it can prove to give excellent results.
 Numenta can also consider for the strategic alliance options with big companies like Google, etc in order to have a partner to achieve the
success in long term. This will also assist Numenta for the financial concerns.
 As Machine learning can dominate the technology completely in the second half of the century and one of our iteration of mission statement
for Numenta was to be catalyst in machine learning. Instead, Numenta created all amazing technology and didn't focus only on being the
catalyst.
Information Classification :Public

THANKS
Group 3