Decision-Making: The New Frontier for Automation

June 2020
Digital Business
Decision-Making:
The New Frontier
for Automation
Decision process automation is a forward-looking, practical
strategy to improve enterprise operations, enabling faster
responses to rapidly changing conditions and identifying
options for action based on a more complete exploration of
potential outcomes.

Executive Summary
Digital Business
2 / Decision-Making: The New Frontier for Automation
Decision-making is a fundamentally human trait. But
today’s businesses must process an astonishing amount of
data to make decisions, even in operational areas such as
supply chain, procurement and contract pricing. Assessing
extraordinary volumes of data is near impossible for even
experienced managers, as they must be mindful of previous
choices and weigh multiple variables that could affect
outcomes.
Through distributed computing and powerful artificial intelligence (AI) and machine
learning (ML), decision process automation (DPA) can address this complexity
by automating how organizations analyze massive volumes of information.
Simultaneously, human experience and insight complement algorithmic decision-
making by assessing contexts too nuanced for AI and ML to address as of yet. DPA
offers a new way to operationalize AI and optimize decision-making, by automating
how organizations analyze what has happened and what is happening to more
accurately predict what could happen.

Digital Business
DPA offers a new way to
operationalize AI and optimize
decision-making, by automating
how organizations analyze what
has happened and what is
happening to more accurately
predict what could happen.
Decision-Making: The New Frontier for Automation / 3

Digital Business
Augmenting the human capacity to solve problems
Across industries, businesses are witnessing massive, unprecedented
growth in the volume of data available for analysis. This avalanche
of data often exists in siloes, which impedes decision-making. If an
organization cannot aggregate, consolidate and correlate data across
siloes and include institutional knowledge such as previous decisions
and their effects, productive insights will be limited and decision-making
constrained.
Moreover, certain types of institutional decisions depend upon particular people. Individual employees or
departments may create models to address the mechanics of their internal processes, such as spreadsheets
or other applications dedicated to a specific task. These models often do not connect to the enterprise’s IT
systems and become tribal knowledge, but rather take the form of conversations or ad hoc collaborations. In
these cases, organizations lose valuable knowledge when roles change or employees depart.
Institutional decision-making depends not only on data inputs but on less easily modeled elements: the
inherent biases of human decision-makers, including people’s inability to recognize mistakes and reluctance
to correct them, policy frameworks and organizational culture. Decisions cannot be made without accounting
and adjusting for these elements—bringing context into the loop.
Decision-making is harder still in uncertain environments created by stressors such as macroeconomic
events, supply chain interruptions or even a global pandemic. Decision-makers often react to uncertainty by
making instinctive decisions, based on previous choices that may or may not have had optimal outcomes.
And each decision affects other steps down the line.
All these complexities invite questions:
❙ How can managers make optimal decisions when they’re overwhelmed with data from multiple sources?
❙ How can organizations collaborate across departmental or geographic lines and ensure they incorporate
relevant institutional knowledge while still being able to avoid biases or previous mistakes?
❙ How can supervisors of critical operational processes understand the ramifications of specific choices, and
decide which ones would result in optimal outcomes?
❙ How can we make decisions at the speed necessary to address rapidly evolving situations, even a black
swan event like the recent pandemic?

Digital Business
The short answer? They cannot.
Facing extraordinary and accelerating data growth, businesses need new ways to quickly analyze the
enormous pool of available data. Incorporating frameworks that reflect institutional culture and behavior is
a step toward informed decision-making that eliminates data siloes, tacitly understood processes, data gaps
and information latency.
Understanding DPA: Generating greater value from available data
Just as technology has created these challenges, technology offers the solution. Using distributed computing,
algorithms that can process massive amounts of data at lightning speed and so-called evolutionary AI,
organizations can continuously model, simulate and choose the best choices for decision-makers—
automatically.
This is the opportunity presented by DPA—the logical next step from robotic process automation (RPA) and
intelligent process automation (IPA) in the evolving continuum of business process automation.
RPA improves operational efficiency while reducing human error. IPA helps organizations automate complex
tasks, leveraging cloud technologies to bind operational systems together. Other functional software
addresses challenges in supply chains and procurement. All are valuable and effective, because they allow
knowledge workers to focus on more value-added tasks. But none of these operational technologies can
aggregate multivariate data, simulate and model the effects of different decisions past or present, or rank-
order optimal decisions based on desired outcomes, allowing managers to then choose one. DPA can.
DPA leverages massive computing power and increasingly sophisticated algorithms to assist in business
process decision-making, with huge potential value.Just as one might create a “digital twin” to model
performance parameters on an aircraft engine, DPA creates a digital model of an organization. It examines
repositories of historical data, trends and decisions, evaluates current conditions by analyzing incoming data
streams from various sources, and models potential scenarios and outcomes.

Digital Business
This is DPA: the capability by which an organization can use an enterprise data analysis platform to create a
virtual image of itself. This allows AI to simulate how different variables affect outcomes and account for policy
frameworks and organizational culture. Using DPA, organizations can better examine evolving opportunities
and formulate previously unseen solutions. Call it a look into potential futures.
DPA moves automation of business processes toward incorporating various types of data inputs and
variable conditions. DPA can capture and retain the elements of inherently loose processes, such as
individual decisions or tacit processes, by extracting and tracking key process details. And by analyzing
a simulated instance of an organization’s past choices and current data inputs, DPA optimizes decision-
making in situations where even the best minds may balk or fall short. It winnows out optimal choices
from overwhelming amounts of data, incorporating algorithms that progressively refine outcomes while
eliminating human biases.
DPA delivers choices based on how the organization has worked in the past even as it incorporates data
streams from the present. This brings humans back into the loop at the most critical juncture in the
decision-making process: where the human capacity for insight and foresight, and the ability to understand
institutional context, offer the most value.
DPA is the future, because it shows organizations what the future can look like—machines present optimal
choices, while humans remain in the decision loop.

DPA’s Building Blocks: RPA
and IPA
Two types of process automation have become pervasive in industry: RPA and IPA.
❙ RPA uses virtual assistants—i.e., software “bots”—to automate routine tasks, which
speeds turnaround times, reduces errors and frees staff to focus on higher-value
work. RPA tends to be localized in operational environments, customized by on-site
process engineers.
❙ IPA typically ties multiple bots together to perform more complex tasks, often between
systems that did not previously interoperate. Using AI, IPA makes sense of larger data
sets to improve processes automatically, with the added benefit of continuous learning
optimization. Algorithms link process automation to specific outcomes based on sets
of variables, offering higher order processing. IPA also reduces demand on human
resources and is often leveraged across geographical units via the cloud.
DPA leverages massive computing power and increasingly sophisticated algorithms to assist
in business process decision-making, promising higher value. In this way, it goes beyond IPA
and RPA by aggregating multivariate data, simulating and modeling the effects of different
decisions past or present and rank-ordering optimal decisions based on desired outcomes,
providing managers with more options.
Quick Take
Digital Business

Digital Business
How DPA works: The power of distributed computing
DPA represents the next step forward in managing enormous data
sets to arrive at optimal decisions. It does not eliminate RPA or IPA, but
uses data generated from both as input. It represents a categorical leap
forward—assisting in decision-making at the supervisorial and process
management levels, even at a strategic level.
Fundamentally, DPA is a higher-order decision engine running in a virtual model of a business’s operations.
It relies on evolutionary AI, running iterative sequences of scenarios, with thousands or even millions of
variables, on a cloud-based network of millions of distributed processors.
This sequential algorithmic analysis allows the decision engine to rapidly refine to optimal outcomes, enabling
automated decision-making at lightning speed. DPA makes choices as humans might, but with the ability to
manage an extraordinary number of complex variables, including data inputs from evolving conditions in real
time, without overlooking data or making decisions based on hunches.
Consider how proactive insight and rapid scenario modeling could help save millions of dollars in contract
pricing annually by repricing contracts for key commodities, raw materials or parts as conditions change,
automatically optimizing the structure and terms of counterparty deals and streamlining customer
interactions. Then, consider how an individual, or even a team of analysts, would struggle to optimize
decision-making in rapidly evolving circumstances.
Figure 1 (next page) shows how DPA can use information streams from tens of thousands of customers to
predict order patterns across products. First, ML solutions analyze variable data from customer contracts
to more accurately predict emerging patterns that could affect contract negotiations. Then, DPA clarifies
at a granular level how contracts are priced and, importantly, how they could be. This produces precise
information—on timing, alternative providers, external forces, or all three—that enables the business to earn a
higher profit per contract while reducing the company’s risk due to financial and market changes.
DPA gives enterprises new, more flexible decision support—the power to evaluate constraints and prescribe
the right spend to the right customer at the right time across every channel, automatically.

DPA: Evolutionary AIs could suggest pricing
based on expectations of futures market
RPA: Automated processes currently reduce
manual steps necessary to fill in contract forms
DPA: Deep learning algorithms could adjust sales
order process to reduce cancellations/revisions
RPA: Manual processes of uploading orders to ERP
system are automated
DPA: Scenario simulations
and event engines can
adjust and optimize
productions plans based
on real time data
RPA: Completion and upload of
invoice forms is simplified through
automation of rote tasks
DPA: ML solutions can utilize data from tens of
thousands of customers to better predict order
patterns across products
Figure 1
Digital Business
Manufacturing/
shipping
R&D
functions
Raw material
inflows
Invoicing
Revenue
recognition
Sales
team
Pricing
contracts
Sales
orders
Supply chain
planning
Robotoc Process
Automation (RPA)
Decision Process
Automation (DPA)
Both RPA and DPAPhysical Flow Information FlowLegend:
Customers

The Power of Evolutionary
Computation
When experts describe the power of digital, they are often alluding to AI’s potential to make
decisions based on data as well as humans or better. But life is not a game of Go; nor does
the world obey a single set of rules.
Evolutionary computation is a powerful new method for leveraging AI across a distributed
network of processors, to apply massive computing power to examine complex, multivariate
problems and drive toward optimal solutions. It borrows from principles drawn from genetic
evolution and mutation, but accelerates the process of evolution exponentially to solve the
most complicated problems and deliver meaningful results.
Evolutionary AI begins by generating candidate “agents” that represent a range of possible
inputs, inflections or mutations, then compares them to distinguish which ones are better
suited to solving a particular problem. Then it runs millions of generations of parallel
algorithmic exploration over a distributed network of CPUs connected via the cloud, vastly
accelerating its arrival at an optimal solution.
ML predicts outcomes with a fixed set of variables by modeling solutions based on past
performance. Evolutionary computation more nearly approximates the human brain,
investigating alternatives based on varying sets of assumptions at a speed and scale
unattainable by humans. It moves AI beyond predicting outcomes to generating complex
models automatically, solving problems creatively and augmenting decision-making by
offering potential new outcomes.
Digital Business
Quick Take

Digital Business
Drive better decisions
DPA can give organizations a deeper view of operations and provide
more clarity in resolving complex problems and predicting outcomes.
It provides a framework to help organizations create higher value—a
decision optimization capability that models different business rules,
eliminates latency in determining what path to take and lessens errors in
decision-making. This supports the organization’s ability to:
❙ Analyze and correlate historical data, details on the outcomes of earlier decisions and the flow of incoming
data from current activities, in order to anticipate future outcomes.
❙ Adapt to changing situations, such as evaluating contract performance, comparing the relative cost of
spot purchases, examining trading and/or hedging decisions, managing supply-chain risk during weather
events, or reducing logistics bottlenecks and costs even in an international crisis such as the recent
pandemic.
❙ Gain insight into what is happening in near-real time, discovering the causes of certain outcomes and
being able to adjust proactively to evolving circumstances, while eliminating human biases.
In the context of supply chains, for example, DPA ML algorithms could perform scenario modeling on
demand, based on inputs about changing conditions. This could lower inventory requirements while
reducing unfilled sales and the number of changes to sales orders. DPA could optimize delivery volumes,
sales routes and vendor selection, increasing on-time delivery while smoothing disruptions. And DPA could
intelligently improve contract pricing, to lessen uncertainty in changing financial conditions and increase
profits per contract.
DPA can build rules based on correlations drawn among the conditions present when a set of decisions were
made, factoring in their downstream impact. Iterative modeling toward outcomes means that once a path
to an outcome is explored, a business can feed process output data continuously back into the DPA engine.
Applied on a larger scale, it can generate insights about the outcomes of historical decisions and their results,
which human decision-makers can use.

Advancing Decision-Making
While Keeping Humans in the
Loop
Organizations today are looking to recover from the swiftest, potentially most damaging
economic downturn in modern history. But many are using outdated, increasingly fragile
models for making decisions in unpredictable circumstances. They can experience a
crippling lag between the time data comes into the organization and when it is processed.
Decision-making variables are dauntingly complex. And while information must be available,
understandable and timely, its sheer volume induces a sort of human overload.
The real cost of decisions can only be addressed by understanding organizational context,
leveraging technologies that learn from previous decisions, modeling human behaviors
and preferences and decision biases to understand how decisions are made, and applying
an institutional framework to create the right heuristics for decision processing. Decision
models need to be refreshed. If an organization wants to realize big gains with automation, it
needs to examine decisions made in everyday processes and at a strategic level.
DPA should be understood as augmenting human capacity, not replacing it. Not every
variable can be modelled or every scenario anticipated. This philosophy should underpin
any effort toward decision automation. Every decision is made with some limitations and
information deficiencies, and human decision-making remains necessary and desirable.
Human ingenuity, our creativity, the sensory perception of what is going on in context and
in the culture and DNA of an organization all demand that humans remain in the decision
loop, even as automated decision processes help correct or guard against biases that can
lead to mistakes or losses.
Digital Business
Quick Take

Revenue growth transformation requirements
It’s not simply about having a relevant offering for your customer, or selling effectively, or marketing effectively, or being able to
execute on transformation in the organization. All of these components need to be integrated for success in revenue growth.
Digital Business
Use case: Driving better decisions in procurement
When automation efforts are begun, processes are measured by how efficiently they are executed, including
whether the tools adopted allow users to complete transactions at an acceptable rate. Businesses can
overlook, however, that simply executing a process faster is not the objective. It is more critical that the
process delivers a high return and a positive impact.
Consider a global procurement group that spends $5 billion annually and wants to meaningfully reduce that
spend.Such a function may employ 100 people or more worldwide to perform processes.Some might talk to
customers to complete purchase orders and place them; others use software to complete transactions.Processes
may formerly have been manual,though repetitive tasks are now performed by robots and automation,from
scanning purchase orders to using optical character recognition (OCR) to assign codes to products.
If the intent of process improvement is to optimize procurement spending by making better decisions in real
time, based on prevailing commodities prices or by changing the amounts ordered from different vendors,then
the question for managers is how to do it when individual agents cannot process all the data needed to make
the best decisions.An automated decision engine can aggregate and analyze spending patterns, identify areas
that can be optimized, and achieve measurable gains by tracking and analyzing the complete process.
Benefits of DPA
A business does not have to revamp its RPA and IPA efforts to adopt DPA.
Information delivered from automated processes at a lower level is vital to
driving the DPA strategy. Bringing DPA online depends on understanding
specific needs for decision-making and where it would have the greatest
impact—to accelerate and optimize organizational responses.
The following conditions could signal such needs:
❙ Existing rules or scenarios are used to address certain situations, and they are repeatable.
❙ A combination of events or environment-related issues occurs repeatedly, and automated rules could
augment human decision-making.
❙ The business encounters scenarios that never happened before and must conceive rules to address them,
while leveraging such new rules to address other scenarios with similar characteristics.

Digital Business
Such conditions suggest asking the following questions:
❙ In what areas could DPA most benefit managers and supervisors? In managing inventory? The supply
chain? Freight management and costing, and ensuring OTD? Adjusting contracts or procurement?
Revenue collection?
❙ Which KPIs correlate to decisions? How could monitoring those KPIs help verify the benefit of DPA? What
is the frequency and volume of decisions?
❙ What applications or systems does the business use to make decisions? Do different departments rely only
on certain systems to make decisions?
❙ What sources of information or knowledge could contribute to better decision-making? What form of
collaboration should be enabled?
❙ Does the business record sets of circumstances and the external conditions of decisions?
❙ At what points do you create, capture and process automation data?
With answers to these questions and the input of an advisor experienced in implementing DPA, the
organization then can create a roadmap to realizing a more agile, accelerated, higher-value decision-making
function.
Learn more about DPA by visiting www.cognizant.com/manufacturing-technology-solutions and completing
the form at the bottom.

About the author
Digital Business
Prasad Satyavolu
Chief Digital Officer & Consulting Leader; Manufacturing, Logistics, Energy & Utilities
Business Unit, Cognizant
Prasad Satyavolu is the Chief Digital Officer and Consulting Leader of Cognizant’s Manufacturing,
Logistics, Energy and Utilities business unit. He is responsible for incubating new solutions,
offerings and commercialization for digital business and advisory services in these industries.
Prasad has written extensively on the future of mobility and energy, connected infrastructure and
manufacturing, monetization and consumer experience. In his nearly three decades of work in the
industry, he has held leadership roles and managed complex business environments. Prasad has
successfully incubated and scaled several business lines, and continues to advise clients on large-scale transformation programs.
He can be reached at Prasad.Satyavolu@cognizant.com.

World Headquarters
500 Frank W. Burr Blvd.
Teaneck, NJ 07666 USA
Phone: +1 201 801 0233
Fax: +1 201 801 0243
Toll Free: +1 888 937 3277
European Headquarters
1 Kingdom Street
Paddington Central
London W2 6BD England
Phone: +44 (0) 20 7297 7600
Fax: +44 (0) 20 7121 0102
India Operations Headquarters
#5/535 Old Mahabalipuram Road
Okkiyam Pettai, Thoraipakkam
Chennai, 600 096 India
Phone: +91 (0) 44 4209 6000
Fax: +91 (0) 44 4209 6060
APAC Headquarters
1 Changi Business Park Crescent,
Plaza 8@CBP # 07-04/05/06,
Tower A, Singapore 486025
Phone: + 65 6812 4051
Fax: + 65 6324 4051
About Cognizant
Cognizant (Nasdaq-100: CTSH) is one of the world’s leading professional services companies, transforming clients’ business, operating and technology
models for the digital era. Our unique industry-based, consultative approach helps clients envision, build and run more innovative and efficient businesses.
Headquartered in the U.S., Cognizant is ranked 194 on the Fortune 500 and is consistently listed among the most admired companies in the world. Learn
how Cognizant helps clients lead with digital at www.cognizant.com or follow us @Cognizant.
© Copyright 2020, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is subject to change without notice. All other trademarks mentioned
herein are the property of their respective owners.
Codex 5467

Decision-Making: The New Frontier for Automation

More Related Content

What's hot (20)

Similar to Decision-Making: The New Frontier for Automation (20)

More from Cognizant (20)

Decision-Making: The New Frontier for Automation