nick brestoff, m.s., j.d. inventor, founder &...
TRANSCRIPT
What’s the pain?
U.S. Commercial Tort Costs = about
$160 billion per year
Towers Watson [NYSE: TW] 2001-2010
Even if an enterprise settles or wins
… it loses!
$350,000-$400,000 per case
So the only winning move is …
Without an early warning of the risks,
GM’s attorneys advised GM’s engineers:
• Instead of “defect,” write “does not meet specifications”
• And sent a PPT: “Some examples of words or phrases that are to be avoided” are: …
Say!
Don’t
“Always, annihilate, apocalyptic, asphyxiating, bad,
Band-Aid, big time, brakes like an X car,
cataclysmic, catastrophic, Challenger, chaotic,
Cobain, condemns, Corvair-like, crippling,
critical, dangerous, deathtrap, debilitating,
decapitating, defect, defective, detonate,
disemboweling, enfeebling, evil, eviscerated,
explode, failed, flawed, genocide, ghastly,
grenade-like, grisly, gruesome, Hindenburg,
Hobbling, Horrific, impaling, inferno,
Kevorkianesque, lacerating, life-threatening ….”
maiming, malicious, mangling, maniacal,
mutilating, never, potentially disfiguring, powder
keg, problem, rolling sarcophagus (tomb or
coffin), safety, safety related, serious,
spontaneous combustion, startling, suffocating,
suicidal, terrifying, Titanic, unstable, widow-
maker, words or phrases with a biblical
connotation, [and] you’re toast.”
—Tom Gara, “The 69 Words You Can’t Use at GM,”
The Wall Street Journal Blog
May 17, 2014
Who’s in the race to monetize Deep Learning?
Apple – Amazon – Baidu -- Facebook –Google – IBM - Intel –
Microsoft – NVIDIA --SalesForce&
We use Deep Learningin our patented (No. 9,552,548) system
to identify Litigation Riskand Provide Early Warning
Deep Learning learns by example -- from external data --
We train our system for a case-type, e.g., discrimination,
using previous discrimination lawsuits
“Learning by example” means:
No key words -- No lists No expert rules
No ontologies – No taxonomies Concept search (LSI or PLSI)? No
Predictive Coding? No
NOS 160 – Stockholders’ SuitsNOS 190 – Contracts
NOS 370 – FraudNOS 410 – Antitrust
NOS 710 – Fair Labor Standards Act
Then you monitor enterprise emails-- internal data --to surface those
risky needles in the haystackSo, as a user, what would you see?
Screen 1 shows the high scoringenterprise emails to you (Enron here)
the riskiest mails to you
Accuracy Scores
43210.5%
With Screen 2 select the emails to read This augments your intelligence
“My employment with Enron
is to be terminated, …”
Then you can access a risky emailin its native (.pst) state
“My employment with Enron
is to be terminated, …”
& start an investigation using yourexisting case management system
“My employment with Enron
is to be terminated, …”
To sum upYou install our system
& monitor enterprise emails-- per policy, i.e.,
no reasonable expectation of privacy --& get an Early Warning
of the risky ones
Notice we’ve preserved the applicability of the attorney
work-product doctrine and the attorney-client privilege
Bottom lineCorporate counsel are closest to the risky data but are blind to the risks -
With this new power, now they can seeDon’t manage lawsuits; avoid them
To get started, see our Demo page
Nick Brestoff, M.S., J.D.inventor, founder & CEO
© 2017 Intraspexion Inc.