© ai-one inc. 2012 © ai-one inc. 2012 Intelligence delivered ai-one ™ Faster Document Review for Legal Professionals with a Personal Intelligent Agent October 2012 BETA
Jul 17, 2015
© ai-one inc. 2012 © ai-one inc. 2012
Intelligence delivered ai-one™
Faster Document Review for Legal Professionals with a
Personal Intelligent Agent
October 2012
BETA
© ai-one inc. 2012
Meet Your New Assistant(s)
You train them,
multiply them,
share them.
No overtime,
no benefits,
no complaints.
© ai-one inc. 2012
• ai-BrainDocs is a personal intelligent agent (software
bot) for finding concepts within documents of any
language.
• Customers are legal, financial and compliance
professionals
• Markets include multi-billion dollar eDiscovery and
eGRCM (Governance, Risk & Compliance)
• First customer shipped
• Early Adopter Version Released October 2012
• Personal (cloud) Version Launch December
Quick Facts
© ai-one inc. 2012
Professionals armed with a personal intelligent agents
they train to identify relevant concepts can save
companies, legal firms and government agencies
massive amounts of time and money.
Big Idea
“digital data growth is explosive and digital data is the stuff of
business and business disputes” - Gartner Magic Quadrant for eDiscovery May 2012
© ai-one inc. 2012
Our solution is the ONLY one built with an ai-one “brain”
(uses ai-Fingerprint technology) that addresses
weaknesses of existing language tools, is language
agnostic, works at the paragraph (concept) level and
derives relevance from the context of use within the
document.
What we do different
“Electronically stored information contains human language, which
challenges computer search tools. These challenges lie in the ambiguity
inherent in human language and tendency of people within networks to
invent their own words or communicate in code.”
- Best Practices Commentary on the Uses of Search and Information Retrieval
Methods in eDiscovery, Sedona Conference
© ai-one inc. 2012
Customer-Problem-Solution
Solution
Personal intelligent agent can read documents to flag those needing review by the professional user
Problem
Documents must be
read by experts and
they don’t have
solutions they can
initiate, train and
launch quickly and
easily. Experts burn
out reading
thousands of similar
documents and
quality suffers
Customer
Expert legal, financial or compliance professional in enterprise or professional services firm
© ai-one inc. 2012
• Relevant document accuracy
• Timeliness- faster project turnaround
• Productivity- review more documents faster
• Higher job satisfaction
• Cost effective on small projects
• Tighter compliance- risk mitigation
• Integration with other eDiscovery processes
Solution Benefits
© ai-one inc. 2012
• Engagement Letters
• Sales/Marketing materials
• PR/8-K events
• Employment Agreements
• Non-disclosure Agreements
• Option Agreements
• Leases
• SEC Filings
• Email and messaging
• Free text in forms
• Social media
Document Types | Processes
• High Volume
• Operations Documents
• Multi-Language (later
release)
• Compliance
• Review & Encoding
• Manuals
• Surveys
© ai-one inc. 2012
paragraph level
concept discovery
ai-one NathanApp
conceptual
fingerprints
personal
intelligent
agents
the brain
the analytics
ai-BrainDocs Intelligence discovered
content library
• compliance
• eDiscovery
Product Overview
we
b
storage
databases
documents
© ai-one inc. 2012
1. Agent(s) defining the concept are created by user loading example
paragraphs for concept “fingerprint”
2. Documents to be analyzed are batched and imported into ai-
BrainDocs case libraries (similar process to indexing).
3. User directs Agent(s) to analyze a library to rank by concept
similarity score
4. User evaluates performance of Agent and continues training or
saves for production
5. Workflow queue is created and tagged documents are processed
6. User (Admin) customizable output
Product Features
© ai-one inc. 2012
Prototype Screen Shot
Input Fields for
creating concept
Agents
Input Fields for
known “always
include” and “never
include” words
Files ranked by
highest concept
score paragraph
Columns
display
document rank
and link to the
paragraph with
highest
similarity score
Export options
© ai-one inc. 2012
• Test runs measured
performance against sparse
vs rich concept definitions
• 200 documents per test
• Docs were sales contracts
• Scores in “rich” case shows
known target docs (black
bars) isolated at top of list
• Dynamic confidence color
bands show user the
improved accuracy as
concept definition is
enriched
Quick, Iterative Train & Test Cycle
© ai-one inc. 2012
Features:
• Concurrent Users
– Batch Processing of Content Library: 1
– Agent Creation: 5
– Concept Similarity Analysis: 5
• Max Number of Documents in Content Library: 1,000 per batch
• Max Number of Agents: No Limits
• Document Types: Microsoft Word, Adobe PDF (readable), Plain Text
Early Adopter (beta) Solution
Hardware Software Operating System
Processor: 1 x Intel Xeon CPU @
2.8 GHz
Memory: 8 GB of RAM
Storage: ~ 30 GB
• OS: ~15 GB
• Application & Server: ~ 5 GB
• Remaining: ~ 10 GB to store
content library (or higher if
necessary)
Microsoft .NET Framework 4
Java SE Runtime Environment Version
7u6 (or higher)
Apache Tomcat Version 7.0.29 (or
higher)
Web Browser:
• Google Chrome v21 (or higher)
• Mozilla Firefox v15 (or higher)
• Internet Explorer v9 (or higher)
Windows 7 64bit
Windows Server 2003 64bit
Windows Server 2008 64bit
© ai-one inc. 2012 © ai-one inc. 2012
If you’re an early adopter of new
technology and want to work with us
to integrate, trial and test ai-
BrainDocs, let’s talk.
Ready now? Give me a call to
setup a demo.
Tom Marsh, COO
ai-one inc. 5711 La Jolla Blvd.,
Bird Rock
La Jolla, CA 92037
Ph: +18585310674
Follow us on Twitter @ai_BrainDocs
Website www.ai-braindocs.com