18+ years of overall experience in IT field. Director of the Engineering. Key skills: - Build highly effective development process (agile delivery model) - Coordinate and enable successful project execution by working with offshore & onsite teams - Team productivity - Scope and risk management - Knowledge sharing framework - Quality process improvement
Report
Хочу поделиться success story, в которой инновационный подход к технологиям победил сильных конкурентов и рассказать, как можно обьединить людей вокруг сложной задачи. A также на этом примере рассказать о взаимодействии групп Sales<->Pre-Sales<->R&D<->Engineering и на выходе получить Customer Success
2. Outline
3
Facts about me
What problems banks are trying to solve?
What prevents them to do that?
Smart Process Automation
Team Collaboration
Q&A
Facts about WorkFusion
3. Facts about me
4
BSU. Faculty of Mechanics and Mathematics. 1995 – 2000
NT lab – 2 years. VHDL developer
IBA – 8 years. Developer -> Team Lead
Exadel – 4 years. Department Manager
Strevus – 3 years. Co-Founder. Director of Engineering
WorkFusion – 1 year. Co-Founder. Director of Engineering
Institute of Technical Cybernetics, National Academy of Sciences
4. Facts about WorkFusion
5
WorkFusion is in 10 NYC Startups That Raised the Most Amount of
Capital in December
WorkFusion Raises $14M to Drive Smart Process Automation in
Enterprise
5. What problems banks are trying to
solve?
6
Reduce cost
Configurability
Automation process
Reduce FTE: 25% or more
6. What prevents them to do that?
7
Tech team can not deliver
Leaving Legacy systems as is
Huge amount of low quality documents
All optimization was considered done by moving to Offshore
Support Chinese, Korean, Japanese, etc.
8. What is “Robotics” and “Cognitive”?
Why are they new?
9
Head work
Hand work Cognitive
Automation
Robotics
“aka” RPA
i.e. entering data from one
application into another
i.e. extracting information from
unstructured documents
Why now?
End of labor arbitrage
+ strong adoption of
self-service across
enterprise
Why now?
Breakaway progress in
AI tech + availability of
data and compute in
cloud
9. 50% impact can be expected from full-stack
implementation of smart automation, as high
as 70% if starting from onshore
10
Initial state Future state
Plain old
offshoring and
outsourcing
Sourcing CognitiveRobotics
10-15%
robotic
automaton on
offshore, 40%
on onshore
resources
10-20%
cognitive
automation on
top of robotics
5-10%
human worker
analytics / UX
improvements
30%
+50%
Smart
automation
Automation of
onshore FTE
remaining due to
regulatory reasons
10. WorkFusion is Smart Automation
11
Human-in-
the loop
Crowdsourcing Statistical
Quality Control
2011
MIT CSAIL lab research leads to R&D
on human-in-the-loop computing
Microtasking Robotics
2012
WorkFusion launches first SaaS platform
for Microtasking in enterprise
Machine
Learning
2013
WorkFusion launches Machine Learning
automation
Smart
Automation
2014
WorkFusion becomes first full stack
robotics + cognitive + human platform
Full stack
Automation
2015
WorkFusion patents Worker Fitness,
Virtual Data Scientist
11. CASE STUDY
Processing of Invoices to extract header information and
individual line-items
12
Situation
A Human Resource software
company processes up to 150k
invoices on a monthly basis
The current processing method is
fully manual, which limits the amount
of information that can be extracted
Approach
Optical Character Recognition
(OCR) capabilities are applied to
turn each invoice into structured text,
which is then passed through a
workflow
Machine Learning models are
applied to automatically extract
values where possible
When human effort is required, a
semi-automated information
extraction task is used to speed-up
the manual work
Impact
FULLY MANUAL
TO
80%
AUTOMATION
1 LINE PER
DOCUMENT TO
ALL DOCUMENT
LINES
KEYING
REPLACED BY
HIGHLIGHTING
AND AUTO-
SELECTION
12. CASE STUDY
Standardizing on processing format to collect key values
from tax documents
Problem
Customer is storing and analyzing
tax documents for its customers. To
add each document to the database
correctly, the Company Name and
Jurisdiction need to be extracted
from the document. The current
process is fully manual due to the
large variety of PDF types and tax
documents
Impact
Utilized WF’s OCR technology to
convert all documents to a format
maintaining context. Configured web
scraper and information extraction
tools to be able to collect data.
Incorporate in a workflow that
incudes normalization of values and
human exception management.
1
13. How does WorkFusion enabled enterprise automation architecture?
14
Enterprise Infrastructure / Enterprise Cloud
Systems of Record (ERP, CRM, …)Data Warehouse / Data Lake
Cognitive Automation (VDS)
Digitization (OCR, Scraping, …)
Workforce Orchestration
AutomationAutomationenablers
Mobile
Business Process Management
Messaging UXDataIntegration(ETL,DQ,MDM)
Robotic Automation (RPA)
AutomationBI/Analytics
EmbeddedAutomation