Extreme Automation

(% MBA 981 – eMasters in FTM @ IITK)

(19th & 20th October, 2024)

Dr. Ramakalyan Ayyagari

Professor of Control Engineering
Dept of Instrumentation & Control Engg.,
NIT Tiruchirappalli
r-k-a-l-y-n@nitt.edu
Course Layout (tentative)
✓Introduction
• 2 lectures (28/09 & 29/09)
• Robotic Process Automation (RPA), AI & Automation, UIPath
• 6 lectures (05 & 06/10, 19 & 20/10, 26 & 27/10)
✓Guest lecture-1
• 1 lecture (13/10)
• Guest lecture-2 & Quiz-1
• 2 lectures (02/11 & 03/11)
• RPM, Case-study, Blue Prism & Other Platforms
• 2 lectures (9/11 & 10/11) + Possible an extra lecture to compensate 12th Oct
• Quiz-2
• 1 lecture (16/11)
• Review of the Course & Closure
• 1 lecture (17/11)
• End Term Examination – As Scheduled by IITK
MBA 981@ eMasters, IITK - Sep-Nov 2024
 Welcome to week: 4 th
th th
19 & 20 October, 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024

Choosing the right RPA platform/tool
• There are several success stories – we have niche players like IBM and
Samsung, visionaries like SAP and PEGASYSTEMS, and leaders like
UiPath, BLUEPRISM, and Microsoft.
• Best way to get a feel for the platform is to download a trial version.
They are great for developer teams to familiarize themselves with the
platform.
• Modern platforms are getting quite intuitive and easy to use, literally
non-technical aka low-code and no-code.
• UiPath, for example, offers free web-based training modules and
knowledge bases. Explore such options.
• But, try not to get caught up in the hype and jargon.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Check if there are additional costs for annual maintenance, for
example, and what level of support is available.
• Licensing structures are also quite diverse. Perhaps, it is better to
choose what suits the project rather than what is, just, economical.
• There exist vendors, Modern Robots for instance, who allow you
to use robots as and when required.
You may please spend an hour-and-a-half here:
https://www.youtube.com/watch?v=oJFfUcqCLO8

MBA 981@ eMasters, IITK - Sep-Nov 2024

Platform Features (generic)
• Modularity
• Turn your robots into modules/layers that can be re-deployed in
similar business areas with minor tweaks
• It could potentially lower the complexity of the project, and also
lower the deployment timelines
• String Operations
• This is most likely the heavily used feature.
• These are operations that permit parsing of text and cycling
through characters of text. [parsing: to examine in a minute way : analyze critically]
• When extracting text, say from an invoice, it may be necessary to
code in the rules for extracting fields such as invoice number from
the correct invoice

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Variable Management
• How does one store data for input, manipulation and output, for
flexibility and security of the platform
• Easy access versus Encrypted variable storage for overall security
of the robot

• Conditions, Loops & Booleans

• In several cases one might see “if-then-else” structures; loops and
flags are also seen
• E.g. UiPath does this using decision boxes, in a flow-chart
representing the workflow

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Security & Error Handling
• Here it refers to the actual security of the platform, thus it is important
• Verifying the standards incorporated, so that the robots do not add
vulnerability to an existing environment’s security
• At scripting stage, robots must NOT be designed to circumvent
controls (literally faithful)
• Accessing/Storing data and Communication among the robots must be
encrypted
• Error handling is also an important element. Scripters can code the
robot to take certain actions in case an error is encountered
• Send an email to an administrator
• Logging data to database for a later troubleshoot
• Follow an alternate path to finish the activity
• Bypass the error and proceed, or Simply stop functioning
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Robotic Platform Security
• It is worth remembering that security related to unpredictability lies in
the way the robots are scripted
• Target environment’s security must be taken into consideration
beforehand
• Some times, say in a bank environment, extra attention must be paid to
the existing security controls; just think of storing passwords within the
robot scripts or centralized server
• In other words, sign-in must be manual; later the robot may take over
• Policies like expiry of passwords after a fixed period of time must
necessarily be captured and accounted for while scripting robots
• Interestingly, on the flip-side, robots could be more trustworthy than
back-office workers
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Robotic Vision
• Vision is one of the senses, and some form of image recognition or
OCR must feature in the RPA development to scrape the data for
processing
• Image recognition comes handy in several other parts as well, click
elements in an application, for instance
• During this data extraction and processing/manipulation, the
physical monitor need not have to be on; [unattended]
• Despite apparent vulnerability, this is of great use for document
text extraction (e.g., paperless loan processing in the back-office)
• Intelligent OCR (ICR)s exist these days to reduce errors of mistaking 0
as 8, and can also “look” at handwritten documents

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Couple this with word recognition, we have IWR (ICR + ML) based
word recognition (say, from a dictionary)
• At the moment there are arguments on either side: ICR/IWR+RPA
solution versus re-engineering and digitizing a process having
traditional hand-written document; but, the former apparently is
gaining momentum
• There is a Citrix workspace (in Google cloud platform) as well for remote
working with a VPN capability.
• In this space, some times it is imperative to look at human
involvement and more accurate devices, or even third-party OCR
engines
• UiPath and BLUEPRISM feature several image-based automation
technologies; e.g., Agentic UiPath

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Cognitive Automation Technology & Smart Process Automation
• SPA, by IRPA, has been defined as a combined approach to
automation that uses RPA with cognitive capabilities
• So far, there have been cases of SPA misused/mistaken as RPA;
terminology isn’t as important as the execution
• However, compared to mere allocation of ‘repetitive’ tasks to robots,
we can think beyond, in the name of SPA/IPA/Advanced PA; in fact,
ever expanding capabilities in the automation space
• BLUEPRISM, for instance, talks about cognitive automation that allows
the use of ML with image recognition or predictive data pattern
technology, basically to allow unstructured document formats to be
read by the robot
• A case in point is the expansion of loan processing where unstructured/
heterogenous documents can also be included, while minimizing the
errors
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Control, Analytics & Insights
• Surge of analytics, with fanciful stories dating close to 2 decades
• There are two contexts that apply to RPA deployments – Analytics
to understand robot performance and robot-related parameters,
and RPA enhanced analytics for converging data sources and
manipulating data (~ inter and intra)
• Most enterprise-grade RPA platforms come with features to track,
monitor and audit your robot activity.
• UiPath, for instance, has a separate module for control center activity
called Orchestrator.
• The basic intent is to have an administrator to monitor, control
and execute robots from this centralized console.
• In Industrialization (cf. groups of robots operating towards
achieving a common business outcome) this comes pretty handy.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• In addition to the analytics, the dashboard will usually show the
status of robots – scheduled, started/killed, errors, transaction
time etc.
• In addition to what is provided by the vendor, the users can also
locally program the robot to store processing information in a
temporary store; these are more useful to troubleshoot at micro
level.
• There exists flexibility to auto-email logs/screenshots to
designated email IDs. [process mining]

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Enhanced Analytics & Insights
• Collection, Sorting, and Funneling of data is usually a very
elaborate exercise, particularly when there are multiple channels.
• Collection itself could be a great challenge.
• Direct database access and APIs provide the analytics platform
data to generate dashboards and allow slicing or dicing of data.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
MBA 981@ eMasters, IITK - Sep-Nov 2024
Bringing Robots to Life
• What environment to use
• Selecting target processes for automation
• Highly manual and repetitive
• Rule based or …
• Low exception rate
• High volumes
• Systems not undergoing major changes
• Robotic Validations
• Automating a Dynamic Process
• Documentation for Robotic Projects
• Calculating automation RoI
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Scripting robots demands a systematic approach. The better your due
diligence and process studies are, the more accurate your robot’s
output is going to be
• Make sure that the development environment features a decent
enough access to the application’s elements
• One of the observations made while deploying robots is that the RPA
project gets treated much like a regular software development
project. Instead,
• Create an environment with dummy accounts and applications,
where we may interact with, but not damage the databases
underlying
• It is essential to figure out options for sandbox environments to
actually perform development and testing
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Any design would commence with requirements and their analysis;
RPA is no exception
• However, since the key advantage of RPA is its speed of
deployment, studies indicate that a lean team that gathers precise
requirements and puts them in code is preferable
• We need to capture all nuances of the process – major pain-points
from managers and owners; how the processes, sub-processes
and metrics are grouped, …

• Off-the-shelf platforms like UiPath, Blue Prism offer non-technical

interfaces
• Combination of drag-and-drop predefined actions trigger the
robot/workflow to be built, encouraging an intuitive “no coding
required” paradigm
MBA 981@ eMasters, IITK - Sep-Nov 2024
Selecting Target Processes
• Obviously, the first aspect is ‘highly’ manual and repetitive process.
• Call Center use case – If we cannot predict what types of calls/mails
are coming into the center, we will find it hard to create robots.
These are dynamic processes, and NLP is a solution
• If the process is defined with a few poorly drafted rules, we need to
elicit and confirm rules with the managers and owners before trying
to explore options to create rules where they don’t exist
• We also need to have a check on exception – too many exceptions
spoil the project! Eliciting tacit knowledge from the employee’s
minds is a big challenge but also the key to understanding
exceptions and figuring out ways to streamline the exceptions and
make the outcomes more predictable
• Perhaps we may even decline RPAing.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Processes with single input type – win over multiple, unstructured
inputs; nevertheless, additional cognitive help is needed for really
complex processes
• Thinking of automation, usually ‘huge volumes’ come to mind;
however, there are no concrete rules to matching a platform to a
process, and a balance among value, security, functionality and
pricing would be the chief criterion
• A trivial example of today is the 3D printer

MBA 981@ eMasters, IITK - Sep-Nov 2024

Changes in the process
• Changes – could be two-fold; regular technology changes in the
process and changes in business logic
• The first one is likely to cause issues, but there are ways out
• The second one has a potentially large impact on the project
• It is worth understanding the client’s future plans
• In the interest of RoI, RPA still makes financial sense even if planned
legacy application upgrades would take a sizeable amount of time.
RPA makes its RoI in parallel while the legacy upgrade is in progress.
• A straight indication – mid- and back-offices, cf. banking, are ripe to
be automated as they are mature. Consequently, we notice that
despite a mish-mash of applications on mainframes and other
isolated databases, it is relatively easy to deploy RPA, to integrate
via robots rather than creating master databases.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Robot’s validation – This is different from a typical human activity;
imagine an officer’s verification of our specimen signature
• When comparing two fields of data in two separate excel sheets, the
robot will usually do a literal matching, and can be programmed to
take action based on the outcome of the comparison; even if the two
fields had a data with potential for variance, the robot will flag the
comparison of the two fields as invalid
• And, the best part of deploying RPA is to address if this validation is
an issue
• Manually program the robot to handle the variations; e.g., we can
create a database of specimen signatures.
• Use a cognitive/AI engine; possibly in a nascent phase now, but
hybrid models have been reported.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• In several cases of hybrid workforce, RPA allows human employees to
locate tasks in a repository and decide between “do it” and “guide
me” modes.
• DO IT button will see the task being performed automatically by
the software.
• GUIDE ME would add on-screen guidance for the steps needed to
be performed by the human agent.
• A savvy RPA architect can re-engineer a process to seamlessly
‘segregate’ manual tasks and robot tasks with a handover between
the human employee and the robot. For instance, a human may
underwrite a loan and email to a specific ID monitored by a robot.
The robot will get triggered by the incoming email to process the
remaining aspects of the task.
MBA 981@ eMasters, IITK - Sep-Nov 2024
Documentation
• Documentation is important – if it is not in the book, it is not there
• We must collect as many SOPs, training manuals and any other
documentation that might tell you about how this process works
• Include all conditions related to the process when documenting the
steps and try to get the descriptions as detailed as possible
• Include exceptions and notes that capture the nuances of the process;
getting a flowchart would be helpful
• Recording a screen while an employee performs the task and adding
a voice-over would also help
• Adobe’s Captivate is one such capture tool

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Documentation is not just at the beginning – it walks through all the
scripting stages ends in the end
• Like any good program, we would liberally annotate to help future
programmers.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Exceptions – Remember they are there
• A note on exceptions – Alas! Not everything can be documented; a lot
of tacit knowledge - experience as you may call it - can be found in
employee’s heads and in order to be successful, we have to find a way
to elicit and capture the unwritten rules of the process
• We may, for instance, interview subject matter experts and
employees working in a process to understand the hidden nuances
hep us capture the missing elements, and subsequently documented
to be converted into automation code
• My own experiences – Fuzzy Logic, Defense’s missiles having
bandwidth problem, Administering exams during pandemic

MBA 981@ eMasters, IITK - Sep-Nov 2024

Managing the RPA project
• Take enough care when designing the RPA project approach
• Make sure we have data backups, and sufficient measures in place
in case the robots malfunction due to incorrect scripting
• Make sure to have a sandbox environment that is distinct from the
live production environment
• A slightly (?) sensitive issue – let the staff continue to work until
the RPA gets stabilized. Not only you need a highly skilled set of
employees to administer the robots, a skilled set must be ready to
work if robots fail
• When the project goes live – track the percentage of automation
calculated at the beginning versus the actual efficiency gained post
automation
MBA 981@ eMasters, IITK - Sep-Nov 2024
Calculating RoI
• It is difficult to templatize, but still the following table might help us:
Factor Using RPA Guiding Metric
# Sub-processes # Automated processes # Un-attended systems

Cycle-time for each process Hours saved Cost/hour saved

Complexity Reduction # Steps/Operations saved

Cost/Error Accuracy achieved # Errors reduced

Human workforce Reduced requirement Salary component saved

Soft savings (May be difficult) Insurance, Training etc. on humans

MBA 981@ eMasters, IITK - Sep-Nov 2024

Post Robot Deployment
• There are reports about companies having their sound technical staff
trained in RPA so that the employees can scope and script robots
• There are certain perception issues though, for a world-wide
acceptance – what if (out of threat, for instance) the result is
poorly coded robots?
• Start-ups may not have a better option
• Bigger companies can go for a CoE to independently handle robot
development
• There are instances reported where (ex-)employees are
incentivized to suggest their own ideas for processes or tasks that
are repetitive, low in value, or rule-based

MBA 981@ eMasters, IITK - Sep-Nov 2024

UiPath & Business Automation
• The UiPath Business Automation platform claims to offer a
comprehensive suite of automation capabilities, covering the entire
spectrum from process discovery and design to automation,
management, and optimization.
• 3 Pillars of the UiPath Business Automation Platform:
• Discover-Automate-Operate.
• Uncovering new possibilities for improvement (Discover),
• Building the end-to-end infrastructure that transforms how you work
(Automate),
• Deploying & managing AI-powered automation across your
enterprise (Operate).

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Claims:
• You can get things done with a digital workforce and automates
work via UI, API, and artificial intelligence.
• Best-in-class UI emulation, powerful API integration and native AI,
like intelligent document processing, all built into the same
platform.
• By using low-code, drag-and-drop tools, every employee can easily
have access to, contribute to, and benefit from automation,
spreading the ROI and benefits to all layers of your company.
• When you automate long-running workflows with seamless robot-
to-human (and back!) handoffs, fragmented processes are
orchestrated.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
Benefits, routinely
• Loan processing time decreased by 60%.
• Customer onboarding time was reduced by 75%, improving customer
satisfaction and accelerating revenue generation.
• Data entry errors saw a 95% reduction, ensuring more accurate
customer records and compliance.
• Monthly reconciliation accuracy increased by 30%, enhancing
financial reporting precision.
• Operational costs decreased by 40% - reduced labour expenses and
fewer error corrections.
• Employee engagement improved as staff focused on meaningful
tasks, increasing job satisfaction and reducing burnout.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• If an insurance company operates across multiple regions, documents
received for policy creation have different languages and formats.
• This lack of standardization results spending a significant amount of
time manually going through the individual documents to extract the
information to create new policies or effect changes to existing ones.
• Automating the onboarding process end-to-end includes creating
new policies and linking them to relevant group policies, calculating
premiums based on the compensation of the policy holder, and
ensuring adherence to various regulations across the regions.
• A detailed assessment, typically suggests, 7-8 processes in the P & C business
for automation. Cognitive OCR and UiPath RPA tools are the right ones to
automate these processes. RPA and Intelligent Automation solutions help in
the complete automation.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• While studying this at home, you may please refer to the following
YouTube video at your own pace:

Rakesh Kumar Behera -

https://www.youtube.com/watch?v=oJFfUcqCLO8

MBA 981@ eMasters, IITK - Sep-Nov 2024

Discover
• Automation Analysts are expected to be familiar with specific tools
and their purposes, to have a good grasp of utilizing tools, and
navigate envisioning future automations.
• Examples include - Task Mining, Process Mining, Communications Mining, and
Automation Hub.
• Task Mining - one can capture and analyze user interaction data with
various applications, systems, and devices to understand how people
perform their work.
• Task Mining helps gain valuable insights into the performance,
identify areas for improvement, and optimize workflows.
• There are two capabilities available under Task Mining - Assisted Task
Mining (ATM) and Unassisted (UTM).
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Assisted Task Mining (ATM) is most beneficial for a Business Analyst
working with an SME on a business process with known tasks.
• Generally, when capturing an “As-Is” process, the Business Analyst
elicits the high-level process flow from an SME during a recorded
video call. The Analyst then documents the human steps into the PDD
(Process Definition Document) based on the recording and his notes.
• By setting up ATM before the meeting, the Business Analyst can
capture the SME's real-time actions (screens, clicks, keystrokes,
hotkeys). The Business Analyst can then analyze, modify, and/or
annotate the flow chart to incorporate into the PDD as needed.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Unassisted Task Mining (UTM) is more suited for the mining of
unknown tasks to discover repetitive tasks for automation.
• In general, data is collected on a pre-selected set of users and apps.
• Over days, the recording runs in the background while the users do
their work.
• Afterward, a machine learning model analyzes the data to suggest
tasks that can be automated.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Some of the other examples of how one can use Task Mining are:

• Process Optimization: To discover bottlenecks and inefficiencies

within business processes, examine user interactions with diverse
applications and systems and pinpoint areas requiring process
enhancement.
• Compliance: Analyse user interactions across various applications,
identifying instances of non-compliance and implementing necessary
corrective actions.
• Employee Training: Identify gaps in knowledge and training
requirements, identify areas where employees face difficulties and
offer tailored training to address these challenges.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Process Mining – one can use data from IT systems to create visual
representations of business processes, and gain insights into process
performance, identify inefficiencies, and optimize processes for
maximum efficiency.
Here are some ways that a Business Analyst can use Process Mining:
• Process Discovery: By analysing event logs and IT system data, one
can use process maps to understand how work is done.
• Performance Analysis: One can identify bottlenecks, inefficiencies,
and other issues that are impacting process performance. They can
make improvements to streamline processes and reduce costs by
analysing execution times, resource usage, and other metrics.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Compliance Monitoring: To monitor process compliance with
regulations and policies, one can identify instances of non-compliance
and take corrective action by analyzing execution data.
• Root Cause Analysis: To perform root cause analysis on process issues
by analyzing execution data and performance, underlying causes.
• Process Improvement: To identify opportunities for process
improvement by analysing execution and performance data, thereby
streamlining processes to reduce costs and improve efficiency.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Discover…
• Automation Hub – one can identify mind-numbing tasks and submit
ideas that will be converted into useful automations; streamline the
process of identifying, prioritizing, and managing automation
opportunities within the organization.
Here is how one can use UiPath Automation Hub.
• Identify automation opportunities and prioritization: By collecting
suggestions for automating processes from different stakeholders
within the organization, one can obtain automation ideas from
different sources and prioritize them according to their potential
impact and feasibility.
• Collaborate with other teams: One can collaborate with users, Subject
Matter Experts, and IT teams and enhance communication and
alignment between the business and technical teams, ensuring that
automation projects meet the organization's needs and goals.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Process assessment and documentation: One can assess the
suitability of processes for automation by documenting the details of
the processes, including their complexity, business rules, and required
inputs and outputs; consequently, one can identify which processes
are most suitable for automation.

• Manage automation projects: Automation Hub provides a centralized

platform to manage the entire lifecycle of automation projects, so
that one can track the progress of automation initiatives, monitor
performance, and ensure timely completion.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Discover…
• Communications Mining – allows organizations to extract insights
from unstructured data in communications, such as emails, chat
messages, and social media posts.
• It uses advanced NLP techniques to analyze and understand the
meaning of text-based communications. Furthermore,
• Customer Service: by identifying areas where they need to provide
additional training or support to their agents.
• Marketing: by identifying areas where the business can improve its
products or services.
• Sales: by identifying most effective sales strategies and tactics,
customer needs and pain points, sales teams tailor their messaging
and offerings to better meet customer needs.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Automate
Studio

• UiPath Studio provides a user-friendly interface that allows us to

create automation workflows using a visual representation of our
business processes.
• One can use it to design and develop automation workflows that
automate repetitive and rule-based tasks in the organization.
• Here are some of the features that are good to know
• Design automation workflows: For example, one can automate the
entry and extraction of data from various sources, such as Excel
spreadsheets, databases, and web applications.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Rapid prototyping: Studio enables you to quickly build prototypes of
automation workflows.
• These prototypes can be tested in collaboration with business users,
validating automation ideas before implementation.

• Data Management and Integration: Using Studio you can orchestrate

end-to-end processes by connecting different software components,
ensuring a seamless flow of data and information throughout the
organization.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Robots

• The UiPath Robot is the execution host that runs processes designed in
Studio.
• Here are some of the features that are good to know.
• Data Entry Automation and report generation: One can extract data
from PDF files, spreadsheets, and web pages.
• Data Analysis: One can extract insights from large datasets, performing
data cleaning and transformation, and creating visualizations.
• This identifies patterns, trends, and insights quickly and accurately.
• Process Automation: One can automate order processing, invoice
processing, and customer service, to reduce manual efforts, increase
efficiency, and improve customer satisfaction.
MBA 981@ eMasters, IITK - Sep-Nov 2024
Action Center
• With UiPath Action Center, you can have a centralized location for
managing and resolving actionable items and user interactions in
automated processes.
Here are a few features that a Business Analyst would find helpful.
• Exception Management: When an automated process encounters an
exception or error, it can trigger an action to UiPath Action Center, to
review and resolve these exceptions.
• Human-in-the-Loop Automation: It allows you to define specific points
in an automated workflow where human input is needed.
• Action Monitoring and Analytics: One can use the monitoring and
reporting capabilities of Action Center to gain insights into the efficiency
of automation, identify areas of improvement, and optimize the overall
workflow.
MBA 981@ eMasters, IITK - Sep-Nov 2024
AI Center
• UiPath AI Center is a service that enables you to build, deploy, and manage
AI models for various business use cases. It also allows you to consume
them within RPA workflows in Studio.
• Here are a few features one would find helpful.
• Build and Train AI Models: One may build and train AI models using a
range of data sources, such as structured and unstructured data, images,
and text to help in automating processes and improving decision-making.
• Deploy AI Models: Once the AI models are trained, you can deploy them to
various environments, such as UiPath Orchestrator or other cloud-based
platforms. This can help to integrate AI models into existing business
workflows and improve efficiency.
• Monitor AI Model Performance: One can monitor the performance of
deployed AI models, including accuracy, response time, and resource
utilization, which in turn facilitates optimization of performance over time.
MBA 981@ eMasters, IITK - Sep-Nov 2024
Document Understanding
• UiPath Document Understanding is an AI-powered feature that can be used
to automate the processing of structured and unstructured documents, such
as invoices, receipts, purchase orders, and more.
Here are a few of the features.
• Automate Document Processing: You can automate the processing of
structured and unstructured documents to reduce manual errors, improve
efficiency, and free up valuable resources for more critical tasks.
• Extract Data from Documents: Using advanced OCR technology, Document
Understanding extracts data from, e.g., tables, forms, and even handwriting.
• Validate Extracted Data: Using the validation feature you can verify the
extracted data against pre-defined rules, business logic, and other validation
checks.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• More on Document Understanding

Mukesh Kala - UiPath Document Understanding Introduction

https://www.youtube.com/watch?v=VHozzZiybJU

MBA 981@ eMasters, IITK - Sep-Nov 2024

Operate
• Let’s focus on Orchestrator and Insights from Operate layer.
Orchestrator
• UiPath Orchestrator serves as a centralized platform for managing,
monitoring, and deploying UiPath robots and automation processes.
• Process management: One can schedule, and monitor processes'
execution, ensuring smooth operation with business requirements.
• Robot Management: One can use the dashboard, and monitor the
status and performance of robots, identifying issues that may arise.
• Reporting and Analytics: A variety of reporting and analytics tools,
provide leverage to identify trends, track key performance indicators
(KPIs), and make data-driven decisions for process optimization.
MBA 981@ eMasters, IITK - Sep-Nov 2024
Insights
• UiPath Insights, offers business intelligence and data visualization
capabilities for the UiPath Automation Platform. One can gain
valuable insights into processes and make data-driven decisions.
• UiPath Insights provides pre-built dashboards that can be easily
customized to help businesses monitor, evaluate, and oversee their
automation initiatives. These dashboards utilize data inherent to
automations created using the RE-Framework's queues and processes
• However, if a client desires the capability to not only view a
dashboard but also delve into specific aspects of transactions,
Business Analysts must identify the extra data fields needed for
reporting in Insights.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• For instance, if a client wants to analyze invoice counts by customer
ID and access individual invoice numbers, additional data
components must be integrated into the data flow.
• By understanding the basic data that is available to Insights from
Orchestrator, one can interview the client to determine whether
additional data elements need to be designed into the automation
processing by the builders. Identifying these custom data elements
before the build phase will avoid the cost of rebuilding and re-testing.
• Here are some features:
• Performance Monitoring: One can actively leverage UiPath Insights to
enable real-time monitoring that allows tracking key performance
indicators (KPIs) such as completed processes, processing time, error
rates, and resource utilization.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Process Analytics: Insights provides detailed analytics on automation
processes, including execution times, frequency, success rates, and
exceptions.
• This helps you to identify variations, understand automation's impact
on efficiency, and uncover patterns or trends that may need attention
or optimization.
• Data Integration: UiPath Insights can integrate with other data
sources like CRM systems, ERP, or databases. By correlating
automation data with customer data, sales data, or financial data, one
can gain a comprehensive view of the impact of automation.
• Compliance and Audit: One can generate reports on user activity,
access control, and process execution details, for compliance
purposes, internal audits, and adherence to regulatory requirements.

MBA 981@ eMasters, IITK - Sep-Nov 2024

UiPath Free Course 1:
• Explore Automation Development with UiPath Studio
(All of you are encouraged to do this course)

MBA 981@ eMasters, IITK - Sep-Nov 2024

Automation Development with UiPath Studio
• The purpose is to provide fundamental knowledge about the UiPath
core components, and get an idea of building processes with Studio.
• A word of caution: Enterprise automation is ultimately software, so it
should be taken seriously. Depending on one’s background, it could
take a while.
• So, what we intended to do is to explore
• How an automation project should look like, how to use the
different blocks and activities, working with the elements in a user
interface, …
• Formulating requirements, understanding roadblocks and
suggesting solutions.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Explorer is the first stage. One is not yet a developer after this, but it’s a
necessary step. And despite prior coding experience, this is still
recommended..
• Next, there would be more learning and some real practice, when one
can easily work on enterprise projects, but more on developing
components, under supervision.
• After some more practice and again some learning about frameworks
and other concepts related to enterprise-scale projects, one becomes
fully autonomous in your work.
• And, since automation is so much more than simple RPA, one can
further specialize in another automation area, like AI, test automation,
or attended automation.
• To begin with, we shall cultivate an educated and technical perspective
on automation, which should help you in your work.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• To quickly summarize, what the Automation Explorer learning path is:
• Identify the UiPath products for end-to-end automation.
• Describe how the UiPath core components (Studio, Orchestrator
and Robot with Assistant) work together.
• Use Studio to create a new project and publish to Orchestrator.
• Use UiPath Assistant to run the newly published process.

MBA 981@ eMasters, IITK - Sep-Nov 2024

What technical setup do we need?
• UiPath Automation Cloud account
• You will need this to:
• Install Studio, Assistant and Robot;
• Distribute attended and unattended licenses;
• Provision robots and run jobs.
• Hardware and software requirements
• Check the documentation link for Studio hardware and software requirements.
• Hardware and Software Requirements - (opens in a new tab) UiPath
Installation and Upgrade Guide
• Check the requirements for installing and running Studio.
• Go to Docs (opens in a new tab)
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Here are a few things to consider:
• UiPath Studio Community will automatically update regardless of your
current version.
• Our recommendation is to install UiPath Studio 2023(4) version.
• Please use the UiPath.System.Activities 2023(4) package version.
• There is a beginner course – Meet UiPath Platform 2021.10

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Studio, Orchestrator and Robots are what we call the core RPA
components of the UiPath Platform.
• We call them the core components because RPA can't work without
developing automations, running them, and managing the entire
ecosystem.
• A quick recap, for all good reasons.
• UiPath Studio is a tool for building automation processes.
• The Robot executes the processes built in Studio published to
Orchestrator or locally.
• UiPath Orchestrator is a web application that acts as the interface
between Studio and Robot and manages, controls, and monitors
automation tasks.
• Let’s see how the core components work together.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• RPA Developers build automation projects using the Studio family, to
fulfil an automation need. These automation projects can then be
published as packages either locally or in Orchestrator.
• Automation projects published to Orchestrator are stored as
packages, along with their versions.
• If an updated version of a project is available, it is stored as a
separate package in Orchestrator.
• To run automation projects, published packages need to be
associated with folders.
• When a published package is paired with a folder, it becomes a
process and can be run by robots with access to the folder.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Human users can trigger robots, through the UiPath Assistant tool to
execute processes on their machines. These are attended robots.
• The robots deployed on separate machines, working without direct
human intervention, are called Unattended Robots. Process execution
for this type of robots is initiated and managed from Orchestrator.
• In both cases, process execution is done through jobs.
• Be it attended or unattended, communication between Robots
and the Orchestrator is always triggered by the Robot.
• This is done by the Robot Service component of the Robot through
a heartbeat mechanism. Every 30 seconds, the Robot Service
sends a heartbeat to the Orchestrator and waits for the
Orchestrator to respond.
• When there is a job to be done, the Orchestrator responds to the
heartbeat by passing the job info to the robot.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• In cases where there are multiple Orchestrators involved in the
architecture, the packages that the robots need to execute jobs are
kept in an external storage location.
• When a robot runs a job for the first time, it will not have the
workflow available locally.
• The robot then requests the workflow from Orchestrator, which
then goes to the storage location, retrieves the correct workflow,
and passes it to the robot. In such cases, even when a Studio user
publishes a package, the Orchestrator moves the package to the
storage location, which is retrieved when required. The robot then
runs the process.

MBA 981@ eMasters, IITK - Sep-Nov 2024

A closer look at each core component
• Studio is an integrated development environment for Automation
Developers to design, develop, and debug automation projects.
• Studio can be connected to Orchestrator, making it easy to publish
automation projects as packages, via the dedicated feeds. From there,
they're distributed to robots to be executed.
• There are two profiles, namely Studio and StudioX. :
• StudioX is meant for business users looking to automate tasks for
themselves and their immediate teams.
• Studio is meant for Automation developers looking to build complex
unattended or attended process automations.
• Depending on the type of Studio license an enterprise has, Studio users
can switch between Studio and StudioX profiles as they need.
MBA 981@ eMasters, IITK - Sep-Nov 2024
MBA 981@ eMasters, IITK - Sep-Nov 2024
• A software robot is an execution agent that runs automations built
with the Studio family and then published as packages.
• There are two types of UiPath robots and they differ both in the way
they work and in the way they're licensed:
Attended robots
• They're digital helpers for human users, and work on the same
machines as humans, during the same hours. They're triggered
directly by humans (usually through UiPath Assistant), or by an event
related to what the human user does. E.g. receiving an email.
• UiPath Assistant is the component that provides a friendly interface
to interact with attended robots. It is the tool that we use to easily
access, manage, and run automations.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
Unattended robots
• These are meant to work non-stop, with as little input from human
users as possible. They're deployed on separate machines and their
jobs are triggered exclusively from Orchestrator.
• Their interactions with human users are typically handled with as
little disruption as possible, by creating and sending requests for
human input or validation as tasks.
• While these await to be processed, unattended robots can continue
their work by picking up other jobs.
• When human input is finally provided, unattended robots can resume
their work on the process.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Automation Cloud Robots
• Automation Cloud Robots (ACRs) are hosted in the UiPath Automation
Cloud that let you quickly run automations without building or
managing your own unattended Robot infrastructure.
• They come in two flavours:
• VM Automation Cloud Robots.
• Access to a fully customizable Windows virtual machine and set up
UiPath Robots to run any job in a matter of minutes.
• Serverless Automation Cloud Robots (Linux based cloud robots).
• Run background, cross-platform projects easily without worrying
about infrastructure setup. So, one can provision, manage and scale
these robots from Automation Cloud.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Test Robots
These robots can be deployed to execute tests on-demand,
continuously, and at scale. Here are some of the characteristics of a
test robot:
• Purpose built for testing workflows, activities, and applications.
• Provides flexibility in managing and reusing test cases across
projects.
• Enables activity level unit testing out-of-the-box.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Orchestrator, the heart of automation management, is a web
application that allows managing, controlling, and monitoring the robots
and the automations.
(Have a look at https://www.youtube.com/watch?v=IuwTbgwINFM)
• One can deploy, trigger, measure, provision, track, and ensure the
security of every robot in the organization.
• It also functions as a repository for libraries, reusable components,
assets, and processes used by robots or by developers.
• However, Building automations and Running automations in attended
mode, are NOT among Orchestrator capabilities.

MBA 981@ eMasters, IITK - Sep-Nov 2024

The main capabilities of Orchestrator are:
• Provisioning: creates and maintains the connection with robots.
• Control and license distribution: enables the creation, assignment
and maintenance of licenses, roles, permissions, groups, and folder
hierarchies.
• Automation storage and distribution: allows the controlled storage
and distribution of automation projects, assets, and credentials, as
well as large files used in automations.
• Running automation jobs in unattended mode: enables the creation
and distribution of automation jobs in various ways, including
through queues and triggers.
• Monitoring: allows monitoring of jobs and robots and stores logs for
auditing and analytics.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Running your first process with UiPath
• Create a UiPath Automation Cloud account.
• Use Orchestrator to invite users and allocate licenses.
• Download and install Studio.
• Use Studio to create a new project, add an activity to the workflow,
and publish to Orchestrator.
• Use UiPath Assistant to run the newly published process.
• Using the integrated core components, the default groups in
Automation Cloud, and the sign-in feature, one can easily set up an
automation development environment.
• To have a good understanding of how the components work, we'll
look at a scenario where a company on a Community plan has
Orchestrator as an Automation Cloud service.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• To learn what all these steps involve, we’ll access UiPath Automation
Cloud to create a new account and organization.
• Afterwards, we’ll invite a new user in the ‘Group Membership’ and
allocate a license type Automation Developer Named User. E.g., we
simply select it from the list and click Save
• Next, we’ll login to Automation Cloud as the new user, download,
install Studio, and create a new process.
• We’ll then add two activities and publish our project to Orchestrator.
• The last step is to run the newly published process with the UiPath
Assistant.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Next, we move to the Orchestrator tab on the left-hand side, select
‘Tenant’, and go to ‘Manage Access’.
•…
• If we chose to stay signed in to Automation Cloud, the web browser
prompts us to open UiPath.
• We continue with Studio and create a new automation process - just
click Start and then Process; We can name it ‘UiPath First Run’, supply
a location, and a description.
• Let’s click Create.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• To start building our automation process, we open the main
workflow.
• For e.g., we may add a simple message box activity that will display a
message on screen when run. We can add activities from the
Activities panel or just by clicking the plus sign in the workflow.
• We can double-click it to add it to our workflow.
• Next, we can type our message directly inside the activity between
quotation marks.
• To save our work , we can press Control + S or just click Save at the
top.
• If we run our process, we can see the message box

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Next, we move on to the Design tab and click Publish to publish our process to
Orchestrator.
• On the ‘Package properties’ tab, we can supply a new package name if needed,
a new version, add a package icon, and so on.
• On the ‘Publish options’ tab, we can specify where we want to publish. We’ll be
publishing this to our Orchestrator Personal Workspace Feed; we can also add a
certificate if needed.
• By clicking Publish, the Info dialog box lets us know that the project was
successfully published.
• Our next step is to run the process by typing UiPath Assistant in the Windows
search bar.
• We can see the newly published process in the Processes list; to run it, click Run.
• Upon running it, the message box is shown on screen.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Here are some of the key takeaways:
• An Automation Cloud account allows a user or robot to authenticate
on your UiPath platform and to receive authorization to access the
resources of your various UiPath products.
• In Orchestrator, on the Manage Access page we can define and assign
roles. We use roles to control the level of access an account should
have.
• The ‘Robot’ role is a mixed role and includes both tenant and folder
permissions.
• In Studio, access the Design tab and click Publish to publish a process
to Orchestrator.
• To Run a process published to Orchestrator, we can open UiPath
Assistant, select the process from the list, then click Run.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• When we publish an automation process to the Orchestrator
Personal Workspace, a process is created automatically if one doesn't
already exist, and existing processes are automatically updated to the
latest published version.
The relation between Automation Cloud and Orchestrator:
• UiPath Automation Cloud is our cloud platform, hosting our cloud
products and services.
• Cloud Orchestrator is a service hosted in Automation Cloud. Users
and groups need to be created first in Automation Cloud. Then you'll
be able to add them and configure their permissions and licenses in
Orchestrator. The Automation Users and Automation Developers
groups are by default added in Orchestrator.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Relation between Attended Robot and Assistant
• Right now, UiPath Assistant is an interface between the human user
and the attended robot.
• It's the default way used to run and customize jobs in attended mode.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Summary
• An Automation Cloud account allows a user or robot to authenticate
on your UiPath Platform and to receive authorization to access the
resources of your various UiPath products.
• In Orchestrator, on the Manage Access page we can define and assign
roles. We use roles to control the level of access an account should
have.
• In Studio, access the Design tab and click Publish to publish a process
to Orchestrator.
• To Run a process published to Orchestrator, we can open UiPath
Assistant, select the process from the list, then click Run.

MBA 981@ eMasters, IITK - Sep-Nov 2024

UiPath Free Course 2:
• Orchestrator Overview for Automation Developers
(All of you are encouraged to do this course)

MBA 981@ eMasters, IITK - Sep-Nov 2024

More About Orchestrator
• We're here to cover Orchestrator from the perspective of an Automation
Developer. We'll focus more on the core Orchestrator concepts and
features, using them in development, and seeing them work at runtime.
• Automations must be thoroughly Discovered, solidly Built, effectively
Managed, reliably Run, and flexibly Engaged with. Not to mention, we
want to accurately Measure & Govern them in all stages of the lifecycle.
• Now it’s time to focus on the Manage and Run phases.
• We can look at an orchestrator this way. Once an automation project is
built and tested, we publish it as packages, and from there we can deploy
it to users’ machines or to robot machines, real or virtual.
• Well, let’s go a little more in depth.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• The automation projects published to Orchestrator are stored as
packages, along with their versions.
• Published packages need to be associated with folders to run
automation projects.
• The pairing of a folder and a published package becomes a process. The
robots, with access to that folder, can then run this process.
• Through UiPath Assistant, human users can trigger robots to execute
processes on their machines. These robots are known as Attended
Robots.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• The robots deployed on separate machines, working without direct
human intervention, are called Unattended Robots.
• Process execution for this type of robot is launched and managed by
the Orchestrator.
• In both cases, process execution is done through jobs. More
specifically, each execution of a process by a robot is called an
individual job.
• Be it attended or unattended, communication between the Robots
and Orchestrator is always initiated by the Robot.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Moving on to the main capabilities of Orchestrator:
• Provisioning: Orchestrator creates and maintains the connection with
robots and attended users.
• Control and license distribution: It also enables the creation, assignment,
and maintenance of licenses, roles, permissions, groups, and folder
hierarchies.
• Running automation jobs in unattended mode: it enables the creation and
distribution of jobs in various ways, including through queues and triggers.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Automation storage and distribution: Enables the controlled storage
and distribution of automation projects, assets, and credentials, as
well as large files used in automations.
• Monitoring: Admins monitor jobs and robots. It also stores logs for
auditing and analytics.
• Inter-connectivity: finally, Orchestrator acts as the centralized point
of communication for third-party solutions or applications.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• A single Orchestrator instance can be split into multiple Tenants.
• Each tenant in an organization can be further subdivided and
organized into Folders.
• Tenants are designed for the purpose of complete isolation of all
Orchestrator entities (i.e., Robots, Assets, Queues, etc.) between the
segregated instances of our deployment, all without having to
maintain multiple Orchestrators.
• Modern folders provide multiple features such as automatic robot
management, hierarchical structures, and fine-grained role
assignment for users.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Consider a typical large company, in which both the data and the
business processes are typically separated between divisions like
Sales and Finance.
• But then, the subdivisions would have some of the data or some of
the processes separated, at the same time, sharing others.
• In Orchestrator, some of the entities exist in the tenant context, while
others exist in the folder context.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• It might seem challenging to use Orchestrator at first, but it becomes
easier once you learn that hierarchy is important, as some entities
exist in the tenant context, while others exist in the folder context,
and that users have different roles and permissions.
• Let’s begin with Tenant-level Entities.
• On the Robots page we have the breakdown of all robots created for
our account with different robot types.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Folders help us maintain fine-grained control over the automations
we publish and their associated resources.
• A Personal Workspace is our own modern folder where one can keep
our processes and assets used in automations, as well as read data
about jobs and logs.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• A machine refers to a physical or virtual machine that's connected to
Orchestrator and has the UiPath Robot software installed on it.
• All of the projects published from UiPath Studio are shown on the
Packages page.
• On the Libraries tab, we can view and delete all libraries published
from UiPath Studio.
• Let's now talk about Folder-level Entities.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• An overview of the entities and activities by folder is available on the
Home page.
• Processes, as we’ve learned, are the automation projects associated
with a folder.
• We can run processes from Orchestrator just by clicking the Start a
Job button!

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Next, Triggers can be used to execute processes in a pre-planned
manner, at regular intervals, when a specific event occurs or
whenever a new item is added to a queue.
• Lastly, the Logs page displays logs generated by Robots when a job is
run. These are useful for debugging purposes.
• Now let's take a look at folder resources.
• Indeed, they refer to entities that can be used in a process, such as
files, folders, credentials, or queues.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Queues are containers that can hold an unlimited number of items.
• Queue items can store multiple types of data, such as invoice
information or customer details.
• This information can be processed in other systems, SAP or
Salesforce, for instance.
• Think of Assets as being specific pieces of information that may
change, such as addresses or values, or that you need to be careful
with, like credentials.
• By storing them here, we add an extra layer of security, and we can
use them for different automation projects.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Storage buckets provide a per-folder storage solution to use when
creating automation projects.
• For example, a .pdf file used across multiple business processes or
data sets used for a machine learning model.
• Next, we understand roles and permissions in Orchestrator.
• In UiPath Orchestrator, roles are groups of permissions that
determine what actions a user can perform within Orchestrator. By
assigning a role to a user, we grant them access to specific features
and functionality based on the permissions associated with that role.
• We can add a Tenant Role or a Folder Role from the Roles tab, which
will provide the necessary permissions for users to access the
features associated with that role.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• As an example, here's a user that has two default roles assigned:
Allow to be Folder Administrator and Allow to be Automation User.
• We can also check this at a Folder level, by going to the Settings page
where we have the Manage Access tab.
• This one displays the accounts currently assigned to this folder. If we
click on Edit for this account, we can see it has the Folder
Administrator role as seen earlier.
• To summarize, folders and tenants allow us to manage and organize
our automation resources in a structured way. Folders provide
isolation at the resource level, while tenants provide isolation at the
organizational level.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Tenant entities
• From the entities defined above, robots are tenant entities.
• This means that they can be allocated to multiple folders in that
tenant. Using roles and permissions, the way robots work with each
of the folders can be customized. We'll see that a bit later.

• Packages are published to Orchestrator using feeds. The feeds can be

configured to be at tenant level, or at folder level.
• A package published to the tenant feed can then be used in a process
in any of the folders. If it is published using a folder feed, it can't be
used for processes in other folders.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• There are a couple of other Orchestrator entities which exist at the
tenant level:
• User – Both human users and robots are uniquely identified with users
in Orchestrator.
• On the Accounts & Groups page, in Automation Cloud, one can define
local user accounts, robot accounts, and local groups.
• The level of access and the actions that your users can perform is
controlled using two elements:
• accounts, which establish the identity of a user and are used to log in
to your UiPath applications
• roles, which are assigned to accounts in order to grant them certain
permissions within the UiPath ecosystem.
• Accounts are not created or managed in Orchestrator, only roles and
their assignments are. MBA 981@ eMasters, IITK - Sep-Nov 2024
• Machine – These are Orchestrator entities corresponding to the
workstations where human users and robots work. Using API keys,
they enable the connection between the physical or virtual machines
and Orchestrator.
• License – The right to use Studio and/or Robots, both attended and
unattended, is done through licenses.
• Licenses exist at tenant level, from where they get distributed to
users, and consumed when the machines connect to Orchestrator.
• Webhook – Webhooks facilitate the communication between
Orchestrator and other applications at API level.
• These are mapped at tenant level, which means they cannot be
differentiated between folders and will provide information for the
entire tenant.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Folder entities
• A folder is a storage area that helps keep your projects separate. From
the entities defined at the beginning of the lesson, processes and jobs
are folder entities. Packages depend on feed configuration.
• Asset – An asset is a piece of data stored in Orchestrator for the use
of robots. There are four types of assets:
• Text - stores only strings (it isn't required to add quotation marks).
• Bool - supports true or false values.
• Integer - stores only whole numbers.
• Credential - contains usernames and passwords that the Robot
requires to execute particular processes, such as login details.
• Assets can have a global value or a value per user. This means that
only the specified user will access a certain value stored in that asset.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Storage Bucket – These are entities used for storing files which can be
used in automation projects.
• Queues – These are containers that can hold an unlimited number of
items, storing different types of data.
• The process of feeding items to a queue is typically different from that
of processing the queue items, and is handled by different robots.
• Triggers – enable the execution of jobs in a pre-planned manner:
• Time triggers—they instruct the automation to start at regular
intervals.
• Queue triggers—they instruct the automation to be activated
whenever new items are added to your queues.
• Event triggers—they instruct the automation to start whenever a
specified event occurs.
MBA 981@ eMasters, IITK - Sep-Nov 2024
• Personal Workspaces
• A personal workspace is a modern folder available for the dedicated
use of a particular attended user. Personal Workspaces make it easy
to deploy automations to your own robot, for easy regular execution,
with the organizational benefits of logging, visibility, and potential
reuse.
• They come with the following entities: package feed, machine
template, and resources (jobs, assets, logs, queues, etc.).

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Roles – are sets of permissions used to control the access of human
users and robots to tenant and folder entities.
• Each permission is defined from the combination of at least an action
type (view, edit, create, and delete) and an entity, be it at the tenant
or folder level.
• For example, you can set up the right to view only the queues in a
certain folder, but not the queues from other folders.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Logs in Orchestrator – are time-stamped files that contain informational
events, errors, and warning messages relevant to the application.
• The UiPath Platform has logging capabilities for all of its main
components. Logs are created locally of every robot and automation
action, then sent to Orchestrator from where they can be filtered,
viewed, and analyzed.
• There are two types of Orchestrator logs:
• Diagnostic logs generated by the Orchestrator regarding its behavior.
• Execution logs are logs generated by process execution. The Logs page
displays logs generated by Robots in all folders you have access to,
including logs generated for jobs started through remote debugging
sessions. To access it, navigate to the Automations tab from a folder
context, and select Logs tab.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Robot provisioning refers to the process of setting up and configuring
robots within Orchestrator.
• You can connect your attended and unattended robots to
Orchestrator in different ways.
• On a high level it involves adding robots to Orchestrator, establishing
the necessary connections, and configuring their properties to enable
effective management and execution of automation processes.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Automatic Robot Creation
• To simplify the attended and unattended robot creation, as well as
the license provisioning, the automation robot creation can be
enabled at user level, for both attended and unattended robots, and
at group level for attended robots.
• Basically, you enable the Attended Robot or Unattended Robot toggle
at the account or group level, configure the various settings (robot
execution settings, machine login credentials, if applicable), and a
floating robot with those attributes is created.
• Note: You can only enable attended robot auto-provisioning for user
groups, unattended robot auto-provisioning is not possible.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Machine Templates
• Robots run on physical or virtual workstations. These are mirrored in
Orchestrator by entities called machines. The machines in
Orchestrator work as API key generators, authorizing the connection
between the robots and Orchestrator.
• There are two types of machines in Orchestrator:
• Machine templates: this allows the connection to multiple
workstations with a single API key.
• Standard machines: this allows the connection between
Orchestrator and a single machine. This is suitable for scenarios in
which robots need to run on specific machines.

MBA 981@ eMasters, IITK - Sep-Nov 2024

License Distribution
• In Orchestrator, licenses are also called runtimes. They're allocated at
machine template level, under Machines in the Tenant menu.
• The number of runtimes allocated there should be matched with the
maximum number of users which can run on a single machine
connected using that machine template. On a regular Windows
machine, only one user can run. But on a Windows server machine,
multiple users can run simultaneously.
• Licenses are consumed as soon as a machine is connected to
Orchestrator, no matter the number of users running on it.

MBA 981@ eMasters, IITK - Sep-Nov 2024

Unattended Automation with Folders
• Functionally, the purpose of attended automation is to have the
robots ready to take over the undesirable tasks when the human
users need it, in their cycle of work and during work hours.

• When it comes to unattended automation, the purpose is quite

different: the robot needs to be busy as much as possible, with as
little human input as possible.
• In what follows we'll first see once again how easily unattended
robots are provisioned. Then we will create and run jobs, making full
use of the license consumption model, process separation based on
UI interaction, and job priorities.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Next, we’ll get to run background and foreground jobs with an
unattended robot that we are going to set up. And while doing all
this, we'll also cover the concept of job priority.
• Before we begin, a word on the background and foreground
processes.
• A background process, also called a headless process, refers to a job
that runs without interacting with any sort of UI. When built in UiPath
Studio, a process needs to be set to start in the background.
• On the other hand, foreground processes use a graphical UI.
• Based on this differentiation, a foreground process can run in parallel
with one or more background processes in attended automation, for
example.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Let’s see how these two processes work in unattended automation
scenarios involving Orchestrator folders.
• The first thing to keep in mind is that unattended robots don't exist as
independent entities. They are a combination between a user or
robot account and a machine.
• To make the setup simple, one setting is enough: enable unattended
robot for the account. This allows that user to use runtimes to
execute unattended jobs on a machine.
• In Orchestrator, we need to define at least one machine template,
which we can then use to connect physical or virtual machines to
Orchestrator.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• What makes unattended automation with folders truly effective?
• Use of Folders and Job Priorities
• By accurately reflecting the business hierarchies with the help of
folders, roles and permissions, we can control the access to
automation processes.
• Job priorities ensure that business priorities are well reflected.
• Processes in Orchestrator are now differentiated between those that
interact with the user interface, the foreground processes, and those
that don't, called background or headless processes.
• This has a significant impact on the way automation jobs are
executed. An unattended robot can simultaneously run a foreground
job and as many headless jobs as available runtimes on a machine.

MBA 981@ eMasters, IITK - Sep-Nov 2024

License allocation per machine
• Allocating licenses (runtimes) per machine makes sure that their
consumption is optimized. For example, on a Windows Server
machine, multiple robots can open sessions and run unattended jobs
up to the maximum number of runtimes.

Custom Job Allocation Strategies

• Unattended automation was designed in Orchestrator to ensure
resource optimization and effectiveness. But there are cases in which
business logic is more important.
• Orchestrator offers a couple of features and options to allow the
customization of the job allocation process so that certain jobs or
resources are available only to certain users.

MBA 981@ eMasters, IITK - Sep-Nov 2024

To continue ….

MBA 981@ eMasters, IITK - Sep-Nov 2024

A Brief Note on Generative AI
[Source: UiPath.com - Five steps to make Generative AI work for your business]

• Just one year on from the public release of ChatGPT, more than a quarter
of Gen X, Gen Z, and Millennials claim to use Generative AI in the office.
• Bill Gates described the technology ‘as fundamental as the creation of the
microprocessor, the PC, the Internet, and the mobile phone’.
• A recent report from Deloitte showed that 79% of corporate leaders expect
Generative AI to transform their organizations in the next three years.
• However, most are focused on the tactical benefits of the technology
rather than using it to drive growth and innovation. Furthermore, only 25%
believe they are well prepared to tackle the risks associated with
Generative AI adoption.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• Previously, automation changed the business landscape through its ability
to reduce operating costs and magnify output across countless industries
and use cases.
• The impact of AI on the business automation landscape is expected to be
even greater.
• AI, working in tandem with automation, represents an enormous
opportunity for executives and C-suite. The new value it can create for a
business is exponential.
• Over the next ten years, Goldman Sachs expects Gen AI to increase global
wealth by 7% (nearly $7 trillion) and boost productivity growth by 1.5%.
• The potential of AI and automation is well-understood. But, actualizing that
potential to deliver real business results has been far more challenging.
• Most executives want to harness AI and automation to increase value for their board
and shareholders. But the business isn’t set up—in terms of enablement or
technology—to take advantage.

MBA 981@ eMasters, IITK - Sep-Nov 2024

1. Understand the potential

• Before you attempt to deploy Generative AI in your business, you

should understand what it is and what it’s capable of.

• Generative AI is a type of artificial intelligence that can create new

content, such as text or images. It does this through a process called
unsupervised learning, where the AI model is fed huge quantities of
data and learns to identify and replicate patterns in the data.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
2. Recognize the limitations

• While it may seem to work as if by magic, it’s crucial to remember that Generative AI is
another tool. And there’s no perfect tool for every job.
• As a developing technology, Generative AI has three main limitations:
Trust and transparency
• The most popular models rely on new data to train and improve, and there’s no
guarantee the data you give to them won’t be shared with another user. What’s more,
people fear what they don’t understand, and there’s much that even the experts don’t
know about Generative AI models and how they work.
• A lack of transparency is part of the problem. These systems are black boxes which take
in data and transform it into something useful through a mysterious process.
• Generative AI can’t be left alone and its outputs must be closely supervised by human
employees in the loop.
• Businesses have a responsibility to protect their customers and a duty to explain their
processes to regulators as well as their shareholders. But Generative AI models can’t
offer clear and understandable explanations of their decision-making processes.

MBA 981@ eMasters, IITK - Sep-Nov 2024

 Context
• ChatGPT and Google’s Bard are best thought of as generalists. They’re
massive models trained on trillions of parameters and data-points. This
makes them useful across countless different use cases, and capable of
answering almost any question you can think of.
• However, while the scope of their knowledge is as wide as the ocean, pry a
little further and you’ll find their understanding is puddle-deep.
• Businesses require accuracy and specificity. Therefore, Generative AI
systems need mountains of business context before they’re ready to be let
loose on your enterprise.
• Without context, a Generative AI system is like having access to a good
writer who doesn’t know anything about your business. They write well,
but you wouldn’t put them on your IT Helpdesk and have them try to solve
customer issues.

MBA 981@ eMasters, IITK - Sep-Nov 2024

 Action
• The fundamental problem remains – by itself, Generative AI can think
but it can’t act. It’s the proverbial brain stuck in a jar.

• “The limitation with Generative AI right now is everyone is treating it

like a curiosity. People are asking it to write poems or asking them
when Van Morrison was born. That’s cool, but it isn’t very useful. It’s
like having a personal assistant who refuses to do anything. Can you
imagine hiring an employee who is only there to talk to you but won’t
do any work?”
Edward Challis, Head of AI Strategy, UiPath

MBA 981@ eMasters, IITK - Sep-Nov 2024

3. Generative AI + Automation

• It is highly likely that Automation is the missing link. Automation feeds Generative
AI context and empowers it to take action. It adds muscle to Generative AI’s
brainpower and gives it the power to change and make valuable impacts.
• Automation-powered AI What makes automation the perfect partner for
Generative AI? It’s integrated across a variety of business processes and can take
real action almost anywhere. Automation feeds Generative AI the context of
multiple data sources, e.g. access to information from critical enterprise systems
such as Salesforce, Oracle, and SAP, as well as websites and social media.
• Whether that’s to personalize customer interactions en masse, or get concise
summaries of complex Know Your Customer (KYC) data to deliver faster decisions,
Automation allows Generative AI to collect all the relevant context it needs, then
gives it the power to act on that data at scale.
• However, this isn’t a one-sided relationship. Automation gives AI the power to
act, but AI in turn gives automation the power to really think and make a big
impact.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• It’s important to remember that Generative AI isn’t the only show in
town. It’s just the latest show, but AI has been around in business for
much longer than that.

• Generative AI generates content for general-purpose use cases, but

businesses have long used Specialized AI models for:
Understanding and interpreting screens – computer vision
Data analysis
Processing documents - intelligent document processing (IDP)
Processing communications – called communications mining

MBA 981@ eMasters, IITK - Sep-Nov 2024

• ChatGPT performs well across a wide number of different tasks and
use cases, but Specialized AI models are more like a perfectionist for a
specific task.

• That’s why it’s an ideal partner, not replacement, for ‘general-

purpose’ generative models, adding accuracy, reliability, and strong
controls to Generative AI’s power and flexibility.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
4. Understand the risks
• Significant questions remain around safety, bias, and data privacy. Over half (52%)
of companies practice some level of responsible AI, but 79% of those say their
implementations are limited in scale and scope. These and other areas need to be
addressed as part of a robust and AI-specific data policy before enterprises can
confidently put Generative AI to work.

• Before you can safely rely on your system, you need to know...
• Who has access to what? We know that AI systems rely on data to continuously improve, but
you need to know your company’s data remains just that. How can you be sure your data
isn’t feeding into the training of third-party LLMs? Could your data be used by competitors to
get an edge in development?
• How secure is your data end-to-end? Where and how is data being sent? How is data
protected at every stage of its journey—when being stored, moved, and processed? What
steps can be taken to protect against data loss and leakage (either malicious or accidental)?
• Are the answers and actions right? Is there bias in the training data that’s affecting outputs?
Are business users making decisions and determinations based on Generative AI
hallucinations? How can we mitigate this? AI success starts with good governance. And only
enterprise-level security guardrails will enable businesses to scale AI-powered automation
and realize Generative AI’s potential—safely as well as successfully.

MBA 981@ eMasters, IITK - Sep-Nov 2024

5. Choose the right cloud provider for Generative AI
• The power of Generative AI plus automation wouldn’t be possible
without the infrastructure to support it.
• The cloud plays a starting role as an enabler of both AI and automation.
• Cost, however, is a big factor: even large enterprises can’t afford to run
the high-performance infrastructure that Generative AI workloads
require.
• But equally important is the access to Specialized AI frameworks, pre-trained
models and APIs.
• By leveraging leading cloud services, businesses can capitalize on
Generative AI without the time and resource commitment, freeing
teams to focus on internal and customer-facing innovations.

MBA 981@ eMasters, IITK - Sep-Nov 2024

MBA 981@ eMasters, IITK - Sep-Nov 2024
Conclusion: The bridge between people and machines
• The value potential of Generative AI is clear. When paired with automation, it
means supercharged productivity, improved data quality, enhanced ROI, greater
efficiency, and the mass customization of customer and user experiences.
• So, what does the future hold? What can we expect as we infuse Generative AI
into business models?
• More specialized systems will emerge: Current Generative AI systems are still very new in
relative terms and they’re understandably extremely general. This will change as businesses
adopt the technology. They won’t need masses of general capabilities, but rather business-
specific, use case-specific systems.
• Machine ‘intelligence’ will grow: We know machines can learn, and this learning potential
brings the very real possibility that they’ll one day have reasoning and perception capabilities
far beyond human scope.
• This will mean faster reaction times and AI that’s refined and customized to every business,
task, and process. It could lead to a whole new operating model for businesses and a
watershed moment in the world of work.
• The cost of adoption will fall: The ability to store models locally and at the edge will bring
down the cost of data processing. This will make it more accessible and encourage the
necessary cycle of experimentation and adoption of AI-powered automation into the
business landscape. Serverless computing will also help improve transparency in AI decision-
making and make hallucination detection easier.

MBA 981@ eMasters, IITK - Sep-Nov 2024

• However, to make this potential reality, Generative AI needs
automation, governance, and the right infrastructure today.
• Automation, combined with Specialized AI, supplies generative models with
crucial business context, the data, and the integration needed to take action.
• Strong governance, enforced by an overarching trust layer, gives you
transparency into how Generative AI models work for our business and
ensure they stay on their best behaviour.
• Finally, your cloud infrastructure is the glue that holds everything together,
letting you capitalize on the best AI innovations for a fraction of the cost of
going it alone.
• Businesses that understand these key ingredients can finally take the
training wheels off and put their Generative AI models to work.
• Please refer to a short video on YouTube on using ChatGPT for UiPath
RPA: https://www.youtube.com/watch?v=6rg6o69AKsI

MBA 981@ eMasters, IITK - Sep-Nov 2024