AI Essentials for HR Professionals

-Krishna Dwivedi
Hello!
Thank you for joining me today

Instructor: Krishna Dwivedi

Microsoft Certified Trainer

Azure AI Engineer Associate AI-102

Azure Data Scientist Associate DP-100

Certified Professional in Neo4j,BCS etc...

Having over 10 years of experience in Training domain

Email: [email protected]
Course Overview and Deliverables
• Why HR should Embrace AI
• Hands-on Learning: Data Science
• Hands-on Learning: Machine Learning
• Exploring Generative AI Tools
• Learn how RPA (Copilot) can help HR to streamline workflows
• End-to-End Project: Generative AI for HR
Why HR Should Embrace The Power
Of AI ?
Drive Organizational Success by Implementing AI in
HR

76% of HR leaders believe that failing to adopt AI within the next 12 to 24 months will
leave their organization lagging in success.

Three-Step Framework for AI Adoption in HR

1.Distinguish Reality from Myth

Collaborate with IT, legal, compliance, and industry experts to debunk myths and
understand the true
potential of AI in HR.

2.Establish the Potential of AI for HR

Align AI use cases with organizational goals, focusing on improving speed, efficiency, and
cost-
effectiveness.

3.Assess and Score Against Success Criteria

Evaluate AI solutions based on governance, workforce readiness, ethics, and risks to make
adoption
 Common Myths vs. Realities About AI in
HR
Myth 1: Tech will replace employees when it performs better than they
do in current tasks.
Reality: AI enhances human capabilities and helps in delegating tasks, becoming
pervasive in employees' activities rather than replacing them.

Myth 2: HR leaders are holding back on AI-powered tool implementation

because of concerns.
• Reality: Despite concerns about data privacy, bias, and ethics, AI adoption is
progressing rapidly. 34% of HR leaders are already exploring generative AI
use cases.
Source: BCG Analysis
What is Artificial Intelligence?
Software that imitates human
capabilities
• Predicting outcomes and
recognizing patterns based on
historic data.
• Recognizing abnormal events
and making decisions.
• Interpreting visual input.
• Understanding language and
engaging in conversations.
• Extracting information from
sources to gain knowledge.
Data Science, Machine Learning, and
AI
Artificial Intelligence
Intelligent software apps and
agents
Machine Learning
Use of data and algorithms to train predictive
models
Data Science
Application of mathematical and statistical techniques to
analyze data
Introduction
to
Data Science
 Data Science
COLLECT DATA

TRANSFORM DATA

ENGINEER FEATURES

DATA VERSIONING
Types of Datasets
• May need to be collected just once, or on a recurring basis.
• One-time analysis of a historical dataset.
• Company's own data.
• Third-party data.
• Ongoing analysis of incoming data, such as:
• Data collected by IoT sensors.
• Sales transaction records.
• System logs.
• May be in different formats, such as:
• JSON
• CSV
• XML
• Relational database/SQL
The Structure of Data

• Data used in machine learning may be:

• Structured
• Facilitates searching, filtering, or extracting data.
• Contained in spreadsheet or database.
• Unstructured
• Not as easy to query.
• Images, video, audio, etc.
• Semi-structured
• Partially conducive to querying.
• Email headers have some structure, even if body of content does not.
• XML/JSON documents may be structured or unstructured.
Unstructured Data Examples

Text Image Video Audio

Terms Describing Portions of Data
Feature

Data example

Value
 Data Quality Issues

• Algorithms perform better when given

large amounts of good data.
• To produce a good model, you must
start with a good training dataset.
• Quality issues you may encounter
during data preparation:
• Irrelevant features
• Non-representative data
• Imbalanced data
• Errors, outliers, and noise
 Data Quantity Issues

1. Refers to number of examples, number of features, or both.

a. Table can have many rows, many columns, or both.

b. Document can have many total words and many different types of words.

2. Different types of quantity can impact machine learning differently.

3. Generally, the more features you have, the better the model.

4. Even with many features, a low number of examples can be a problem.

a. E.g., 80 columns but only 10 rows.

b. Model may not be able to make a good estimation.

c. Large number of examples can minimize effects of bad data.

5. Not always feasible to have a high-quantity dataset.

6. Rule of thumb: At least 10 times as many records as features.

Data Science Lifecycle

10/23/2024
Exploratory Data Analysis
Ethical Considerations in Data
Collection
• PII (Personally Identifiable Information): PII refers to data that can
identify individuals, allowing them to be contacted or located without their
consent.

• Data Usage: People often want to know how their data will be used upfront,
even though usage may change over time.

• Data Quantity: While more data is beneficial for models, it must align with
privacy and security considerations.

• Data Bias and Intellectual Property: Biased data can harm individuals or
groups, and collecting data without consent may violate intellectual property
rights and lead to legal issues.
Transform Data
Types of Data
Feature types:

a. Quantitative

i. Number as a magnitude, e.g., Miles Driven.

b. Qualitative

i. Non-ranking categories, e.g., Vehicle Type.

c. Ordinal

i. Ordered data, e.g., Condition.

Data types:

d. Integers (25)

e. Floats (12.78)

f. Strings ('Sedan')

g. Booleans (True or False)

h. Datetimes (2022-12-01)
Data Cleaning
Data cleaning: The process of addressing inaccuracies and other problems with
data.

• Subset of preparation.
• Duplicated data, poorly formatted data, corrupt data, etc.
• You can correct data or remove it.
• Choice of action depends on feasibility and impact on later processes.
Operations You Can Perform on Different Types of Features

Feature Type Description

Dataset: (18, 18, 29, 45, 59)

• Mean: (18 + 18 + 29 + 45 + 59) / 5 = 33.8
Quantitative • Median: 29
• Mode: 18
• Range: 59 − 18 = 41

Dataset: ('Sedan', 'Pickup truck', 'SUV', 'SUV', 'SUV')

Qualitative
• Mode: 'SUV'

Dataset: ('Excellent', 'Very good', 'Good', 'Fair', 'Fair')

Ordinal • Median: 'Good'
• Mode: 'Fair'
Missing Values

1. You should actively decide how to handle missing values.

2. Sometimes models can cope with missing values, other times not.

3. You may drop entire records with missing values.

a. Can reduce effectiveness.

4. You may use a best guess or arbitrary default to fill in missing values.

a. Simplest and least effective approach.

b. E.g., filling in Rs10000 for missing smartphone MSRP.

5. Imputation fills in missing values using best estimates.

Feature Engineering
Feature Scaling
Feature scaling: The task of transforming the values of multiple features so that
those values are all on a similar scale.

• Distribution of data is sometimes most important factor.

• Scaling functions transform numeric variables to emphasize distribution.

• De-emphasizes magnitude of values.
• Important for distance-based algorithms.

• Example features: Salary and Years of Service

• Former can extend into the hundreds of thousands.
• Latter only extends into the tens.
• Both equally useful for determining employee churn.
• Algorithms might treat former as being more important.
• Scale them so that their distributions are emphasized.
 Normalization

𝑥 − min ( 𝑋 ) Normalization—The technique of transforming a feature so that its lowest value is 0 and its highest value is 1.
𝑥′ =
max ( 𝑋 ) − min ( 𝑋 ) Where:
• x is the initial value being normalized.
• x' is the new normalized value.
• min(X) is the smallest value in the feature that x is taken from.
• max(X) is the largest value in the feature that x is taken from.

• Algorithm sees Salary and Years of Service on same scale.

• Useful with algorithms that don't assume specific distribution.

• Good at minimizing effect of outliers.

Dimensionality Reduction
Dimensionality reduction: The process of simplifying a dataset by eliminating
redundant or irrelevant features.

Feature selection:
Selects subset of original features.
Includes most relevant, excludes most irrelevant.
Useful when there are many more features than examples.

Feature extraction:
Derives new features from original features.
Combines multiple correlated features into one.
Useful in computer vision applications.

Dimensionality reduction is almost always worth doing.

Feature Correlation Analysis
Correlation is a statistical measure that expresses the extent to which two variables
are linearly related. It ranges from -1 to 1

• Correlation analysis helps to determine which relationship between features.

• Strong Positive Correlation +1: When one feature increases, the other feature also
increases (e.g., Salary and Age).
• Strong Negative Correlation -1: When one feature increases, the other decreases (e.g.,
Attrition and Performance Score).
• No Correlation 0: No apparent linear relationship between the features (e.g., Gender and
Salary).

• Use correlation analysis to identify redundant features

• To calculate the correlation between two columns in Excel, you can use the CORREL function
=CORREL(array1, array2)
Activity: Transforming Data
• You will be exploring the employee Dataset.
• Now you'll apply transformation techniques, including:
• Data type conversion
• Deduplication
• Missing data imputation
• Dropping outliers
• Finding Correlation
Knowledge Check
Which of the following is considered a quality issue in a dataset?
A) Having a large amount of relevant features
B) Using representative data
C) Having a balanced distribution of classes
D) Including irrelevant features

What type of data issue occurs when one class is significantly underrepresented
compared to others?
A) Irrelevant data
B) Non-representative data
C) Imbalanced data
D) Noisy data
Machine Learning
 Types of machine learning
Machine
Learning

Supervised machine learning Unsupervised machine learning

Training data includes known labels Training data is unlabeled

Regression Classification Clustering

Label is a numeric value Label is a categorization (or class) Similar items are grouped togethe

5
Binary classification Multiclass classification
Label is or is not a class Label is one of multiple classes
Predict the number of
ice creams sold based Separate plants into groups
on day, season, and based on common
weather Predict whether a patient is Predict the species of a characteristics
at-risk for diabetes based on penguin based on its
clinical data measurements
What is machine learning?
Creating predictive models by finding relationships in data
1. Training data 2. Algorithm 3. Model
(past observations) (Generalizes the (encapsulates the
[x1,x2,x3], y relationship between x and function)
Training

[x1,x2,x3], y y as a function)

[x1,x2,x3], y
[x1,x2,x3], y y= f(x)
[x1,x2,x3], y
Features Label
(x) (y)
Inferencing

[x1,x2,x3] ŷ
4. Inferencing Prediction
data (inferred label)
(unlabeled features)
Model training and evaluation

Multiple training iterations with different algorithms and parameters

1. Training data 2. Algorithm 3. Model

(past observations) (Generalizes the (encapsulates the
[x1,x2,x3], y relationship between x and function)
[x1,x2,x3], y y as a function)

[x1,x2,x3], y
[x1,x2,x3], y y= f(x)
[x1,x2,x3], y 5. Evaluate
[x1,x2,x3], y model (compare
[x1,x2,x3], y predicted labels to
[x1,x2,x3], y 4. Predictions actual labels)
(from the validation
Validation data
data)
(with known labels)
Predictive
Analytics in
HR
• Predictive analytics can
help HR departments
identify trends and
patterns in employee data

• Predictive analytics can

lead to more proactive
decision-making and
better outcomes for both
employers and employees
Activity: Building
Machine Learning in
Excel
Demo + Activity :
Artificial Intelligence
AI: Definition and Types
Artificial Narrow Intelligence (ANI)
Artificial narrow intelligence (ANI) refers to machines that are designed for
specific tasks, such as voice recognition or playing chess. ANI is the most
common type of AI used today.

Artificial General Intelligence (AGI)

Artificial general intelligence (AGI) refers to machines that can perform any
intellectual task that a human can. AGI is still a developing technology and
has not yet been fully realized.

Artificial Superintelligence (ASI)

Artificial superintelligence (ASI) refers to machines that are more intelligent
than humans. This is a theoretical concept that has not yet been achieved.
Deep learning

Human neural network Artificial neural network

x w

∫(f (x,w))
• Neurons fire in response to • Each neuron is a function that operates on
electrochemical stimuli an input value (x) and a weight (w)
• When fired, the signal is passed to • The function is wrapped in an activation
connected neurons function that determines whether to pass
the output on
Role of AI in HR
Recruitment
AI can assist with recruitment by automating tasks such as resume screening, candidate
matching, and interview scheduling, freeing up recruiters' time to focus on higher-value
tasks.

Employee Engagement
AI can help improve employee engagement by analyzing employee data to identify
areas of improvement, such as training needs or communication gaps, and making
personalized recommendations to employees.

Performance Management
AI can provide valuable insights into employee performance, such as identifying areas of
strength and weakness, and making data-driven recommendations for improvement.
Fun Activity: Building Image
Classification Model
Demonstration + Activity
Fundamentals
of
Generative AI
What is generative AI?
1950 Artificial Intelligence
s the field of computer science that seeks to create
Artificial Intelligence intelligent machines that can replicate or exceed
human intelligence

1990
Machine Learning s
Machine Learning
subset of AI that enables machines to learn from
existing data and improve upon that data to make
decisions or predictions
Deep Learning 2010
s Deep Learning
a machine learning technique in which layers of
neural networks are used to process data and
make decisions
2021 Generative AI
Generative AI Create new written, visual, and auditory content
given prompts or existing data.

© Copyright Microsoft Corporation. All rights reserved.

The Journey
to
Generative
AI
What is Generative AI?
Language model

Natural
Image
language Code generation
generation
generation

Write a cover letter for Create a logo for a Write Python code to
a job application florist business add two numbers

Dear <recipient>, def add(n1, n2):

Please find enclosed result = n1 + n2;
my application for the return result;
role of …
Top AI Platforms
Microsoft Co-pilot

Microsoft
Copilot
Copilots
What are copilots?

Generative AI powered assistants integrated

into applications, often as chat interfaces

• Copilots provide contextual assistance with

everyday tasks; typically using a language
model to interpret and generate natural
language
• Business users can use copilots to boost their
productivity and creativity with AI-generated
content and automation of tasks
• Developers can:
• Extend copilots by creating plug-ins that
integrate them into business processes
and data
• Create custom copilots to build generative
AI capabilities into apps and services
Copilots
Levels of copilot adoption

Unlock productivity across your business with Microsoft Copilot

• Use copilot to optimize the way you work in the tools you use every day
• Empower employees to be more creative
• Spend less time on mundane tasks and more on high impact activities

Extend Microsoft Copilot for your business with plug-ins

• Integrate business-specific tasks and processes into Microsoft 365
• Get actionable insights from your own data and systems in productivity apps
• Build 3rd-party plug-ins so your customers can use your services in Microsoft
365

Build custom copilots of your own

• Integrate generative AI into custom workloads and apps
• Create compelling customer experiences and commercial products
• Have full control over design and development
Considerations for copilot prompts

System Message:
Summarize the key You are a helpful assistant…
considerations for adopting +
Copilot1 described in this Conversation history5:
document2 for a corporate User: Hello
executive3. Format the Assistant: Hi. How can I help?… Language
summary as no more than Copilot + Model
Current prompt:
six bullet points with a User: Summarize the key …
professional tone4. (May be further optimized by Copilot)

1. Start with a specific goal for what you want the copilot to do
2. Provide a source to ground the response in a specific scope of information
3. Add context to maximize response appropriateness and relevance
4. Set clear expectations for the response
5. Iterate based on previous prompts and responses to refine the result
Co-pilot generation capabilities DALL E
Generative AI models can edit and create images. The model that works with images is
called DALL-E, which supports image creation, image editing and image variations
creation.
Image generation: With DALL-E, you can even request an image in a particular style.
Styles can be used for edits and variations as well.
 Editing an image: DALL-E can edit the image as requested by changing its style, adding or
removing items, or generating new content to add.
 Image variations: Image variations can be created by providing an image and specifying
how many variations of the image you would like.

Prompt: Create four

variations of an image of
an elephant with a
hamburger.
Google Gemini?

Gemini, Google's advanced AI model,

excels in text, image, video, and audio
processing.

It's a versatile multimodal model

proficient in complex tasks across
mathematics, physics, and more.

Additionally, it can generate high-quality

code in various programming languages.

Gemini
 Demo on
Copilot & Google AI
Studio
Data Analysis with Generative AI

• Generative AI can create valuable insights from large and complex datasets.

• Advanced models enable accurate predictions, even with incomplete or

unstructured data.

• Automated data analysis frees up time for interpretation, decision-making, and

action.

• Watch Video
Activity:
Data Analysis on Google AI
Studio
Knowledge check
1 What are Large Language Models?
⃣Models that only work with one language.
⃣Models that only work with small amounts of data.
⃣Models that can interpret and generate natural language.

What is Generative AI primarily used for?

2
⃣Data entry and processing
⃣Network security and firewall management
⃣Creating new content such as text, images, and music

3 Which of the following is an example of a generative AI tool?

⃣Microsoft Excel
⃣Google Chrome
⃣ChatGPT
Prompt Engineering
Prompt Engineering – Prompt Elements
O F G R E AT P R O M P T S F O R C O M M U N I C AT I O N S

PERSON PARAMETER
A S
Tell the tool its job title (e.g. its role) Tone, style, length
• Internal communicator • Casual, upbeat, professional,
• Program manager clear and simple, 6th grade reading level
• Storyteller • Short, 1500 characters, 20-minute
presentation
OBJECTIV CONTEX
E T
What do you want it to do? What points should be covered? What
• Write an action required email is the call to action?
• Write a project status update • Direct readers to complete the following
• Write an outline for a persuasive speech action [include CTA]
• Follow this outline [provide bullets]
• Highlight the following features [list
AUDIENC
features]
E • Include these key points [include points]
Who will receive your message?
• Microsoft employees
• Customers
• New hires
• Executives
Primary, supporting, and grounding content
---
Section markers
(--- or ###) Thanks for your interest in our summer school program.
We’re looking for your ideas for webinars on topics such as AI,
web development, … <more…>

Primary content to be ---

summarized, translated, etc.
Topics I'm very interested in: AI, webinar dates, submission
deadlines

Supporting content to Extract the key points from the above email, and put them in a
provide clarity / specificity bulleted list
---
Artificial Intelligence (AI) has evolved over many years …
Grounding content to <more…>
define scope for questions
---

Where and when did the field of AI start?

© Copyright Microsoft Corporation. All rights reserved.
Requesting output composition

Write a table in markdown with 6 animals in it, with their genus and species

| Animal | Genus | Species |

| ---------------- |:--------------:|:--------------:|
| Red Panda | Ailurus | fulgens |
| African Elephant | Loxodonta | africana |
| Snow Leopard | Panthera | uncia |
| Arctic Fox | Vulpes | lagopus |
| Chimpanzee | Pan | troglodytes |
| Siberian Tiger | Panthera | tigris altaica |

© Copyright Microsoft Corporation. All rights reserved.

Knowledge check
1 What are Large Language Models?
⃣Models that only work with one language.
⃣Models that only work with small amounts of data.
⃣Models that can interpret and generate natural language.

What is the main purpose of using examples in a prompt for a generative AI model?
2
⃣To test the model's memory
⃣To increase the length of the prompt
⃣To provide context and guide the model's responses

3 How can you reduce bias in the outputs generated by a prompt?

⃣By including diverse examples and viewpoints

⃣By making the prompt as vague as possible
⃣By using a single perspective in the prompt
Robotic Process
Automation (RPA)
Introduction to RPA
Definition: RPA automates repetitive, rule-
based tasks using software robots (bots).

Key Features:
• Mimics human actions.
• Works with existing software without human
intervention.

Benefits: Reduces manual effort, improves

accuracy, and increases efficiency.
The Importance of RPA in HR

Challenges in RPA Solution:

HR: High volume Automates routine
of repetitive tasks, tasks, enabling HR
time constraints, professionals to
and the need for focus on strategic
accuracy. initiatives.
Applications of RPA in HR
Recruitment:
• Automating resume screening.
• Scheduling interviews automatically.

Payroll Processing:
• Automates payroll calculations and disbursements.

Employee Onboarding:
• Automates document verification.
• Assigns training automatically.
Data Privacy and Security
• Protecting employee data is a top priority for
companies
• Data privacy and security measures must be
in place to safeguard information
• Ethical use of data ensures fair treatment of
all employees
Bias and Fairness in AI
Models
• AI models can contain biases that unfairly
impact employees
• Steps must be taken to identify and address
biases in AI algorithms
• Fair treatment of all employees is essential
for a positive work environment
 Activity: Identifying Bias and Fairness
in Gen AI
Objective:
1)Identify and analyze both text and images generated by Generative AI (Gen-AI) to
uncover potential biases and discuss fairness.
Total Activity Time : 30 mins

Step 1 : Group Formation: Form groups of 3 to 4 participants.

Step 2: Text Generation Prompt on Job Description:

1) Write a job description for a customer service representative.
2) Describe the ideal candidate for a marketing manager position.

Image Generation Prompt on Workplace Diversity:

1) Generate an image of a diverse team working together in an office.
2) Create an image of a company event celebrating different cultures.

Analysis and Discussion:

• Each group will review the AI-generated responses and identify any potential biases.
Consider aspects such as gender, race, age, and other stereotypes.
• Groups will discuss their findings and note any patterns or recurring biases.
Transparency and
Accountability in AI
• Transparency is essential for building trust in
AI decision-making processes
• Companies must be accountable for
decisions made using AI models
• Clear communication with employees about
AI usage fosters a positive work
environment
 Copilot for Microsoft 365: HR Use Case
Copilot for Microsoft 365 is a powerful tool that can significantly enhance the efficiency
and effectiveness of Human Resources (HR) here are some ways it can be beneficial:
1.Streamlined Recruitment:
a.Copilot automates resume screening, making it easier to identify top talent.
b.It assists in creating job descriptions, ensuring clarity and alignment with
organizational needs.
2.Efficient Onboarding:
a.Simplifies document management during the onboarding process.
b.Provides personalized onboarding plans for new hires, ensuring a smooth
transition.
3.Enhanced Employee Engagement:
a.Analyzes employee feedback to provide valuable insights.
b.Recommends tailored training programs to boost engagement and productivity.
4.Data-Driven Decision-Making:
a.Leverages Copilot’s big data platform to make informed HR decisions.
b.Enables evidence-based strategies for workforce management.
Copilot value journey
Grow value from a base of solid usage and individual productivity by optimizing
functional processes and impacting the metrics that are important to your
organization.

Generate synergies between departments

Line of Business Copilots Measured as revenue gains and cost reductions
Org Business Copilot extended into
(Close rate, Onboarding time, Supply chain
Extensibility process flows
Value costs, Outside legal counsel costs, IT
development costs, etc.)

Role-specific prompts Extend usage to role-based processes

Functional Functional
Multi-turn prompts Measured as KPIs (e.g. leads pursued for Sales,
Value Skills candidates interviewed for HR, etc.) Can also
Copilot Studio have goals based on usage.

Copilot functions and

surfaces Inspire quick wins to reach value tipping point
Genera
Individual Basic prompts (Meeting Measured as usage and time savings with goals
l Recap,
Value Skills
to improve job effectiveness, work capacity,
Summarize an email, etc.) and employee retention.
Copilot Lab
Top 10 to "Try First"
with Copilot for Microsoft 365
Foundational skills for new users

1 2 3 4 5
Recap a meeting Summarize an Draft email Summarize a Tell me about
– let Copilot keep track of email thread – personalize the tone document a
key topics and action items – get quickly caught up to and length. – get right down to topic/project
so you can stay focused a long, complex email business by summarizing
during the meeting and – provide insights and
thread. long documents and analysis from across
avoid listening to the focusing
recording after. Draft an email to [name] multiple sources to
that informs them that on the relevant sections. getTell
upmeto what's
speednew about topic
Draft an email with notes Click on the Summarize Give me a bulleted list
Project X is delayed two
and action items from icon of key points from file quickly.
organized by emails, chats, and
weeks. Make it short and
meeting files?
casual in tone.

6 7 8 9 10
Give me some Help me write … What did they Revise this Translate a
ideas for … – jumpstart creativity and say … content message
– boost your creativity write and edit like a pro by – when you vaguely – when you’ve got a – with business becoming
with ideas for your work getting a first draft in remember someone rough draft of an idea, increasingly international,
such as agendas, seconds. mentioning a topic, have turn it into usable text it’s important to be able to
product names, social Copilot do the research. and then vary the read or write messages in
media posts, etc. length and tone. other languages.
Suggest 10 compelling Generate three What did person Rewrite with Copilot Translate the
taglines based on file ways to say [x] say about topic following text into
French:

For more prompts, visit Copilot Lab at: aka.ms/CopilotLab

”Effort” level for each scenario

Buy Extend Build

1 2 3

Tailor your data, systems and

Unlock productivity out of the box Create managed copilots with a guided
workflows via Copilot Studio
low-code experience
extensions
• Microsoft Copilot Copilot Studio Custom Copilots
• Copilot for Microsoft 365 • Actions (plugins) Power Automate & Power Apps
• Copilot Role-based • Knowledge (via connectors)
extensions Create copilots with custom code
and full control with pro code
Teams Toolkit in Visual Studio
Azure AI Studio and Microsoft Fabric
Copilot
scenarios for
HR
 Copilot scenarios for
HR

Overview and KPIs Use case by role Day in the life

KPIs play a crucial role in Copilot can simplify the tasks See how real-life HR
organizations, providing a that HR employees perform employees are using Copilot in
compass to navigate toward every day. Look at key use their day-to-day.
success. Let's dive into KPIs for cases and how Copilot can be
Human Resources and how your AI assistant along the
Copilot can assist. way.
Using Copilot in HR Transform HR processes
With changing worker expectations, multi-
generational workforces, and hybrid workplaces it’s
becoming harder to keep talented people. Copilot
Goals and
can help to simplify access to data, improve quality Compensation Employee Learning &
challenges and increase creativity, resulting in improvements & benefits engagemen development
to essential key performance indicators. t

Copilot helps you improve HR processes and access to Microsoft Copilot opportunity to impact key departmental
information. KPIs
• Streamline your recruiting process Organizations can improve the hiring process by using Copilot
Copilot can • Managing internal job transitions to help write job descriptions, summarize interviews, avoid bias
Cost per hir
assist • Streamline benefits and compensation
e when identifying top candidates, improve candidate
with ... • Improve organizational health metrics communications, and improve onboarding processes.
• Improve onboarding and development processes
• Candidate Search
Copilot can help all employees get more satisfaction in their job
• Deliver insights to managers by assisting with challenging tasks and instead working on
• Resolving employee issues
Employee tur
nover rate higher value activities. It also makes many tasks faster, which
• Update policy document can reduce overtime requirements.
• Analyze hiring practices

Speed time to value of new employees by simplifying access

Onboarding ti
me to information and training.

HR roles Employee Revenue growth Cost savings and avoidance Improve Employee Experience
HR Planning and Talent Improve employee
Developmen Revenue growth Cost savings and avoidance
Compliance Business Acquisition experience
t Alignment
 Available Scenario
HR | Streamline your recruiting process with:
Microsoft Copilot level: Buy
(Microsoft Copilot only)1
Reduce onboarding Employee
KPIs impacted Cost per hire Value benefit Cost Savings
time experience

1. Create a Job Description 2. Create Interview Questions 3. Research salary levels

Ask Microsoft Copilot to create job Brainstorm interview questions specific for Find typical salary levels for positions based
descriptions any role with Copilot. Create both behavioral on geography.
for any role. and technical questions depending on your
needs.

Copilot Copilot Copilot

Prompt: Write a job description for a Head Prompt: Generate 5 interview questions for Prompt: Provide a typical salary range for a
of Product Director role with 5+ years of an Executive Assistant with 10+ years of Head of Product Director role with 5+ years
experience. experience. of experience in these five cities.

5. Create Onboarding
6. Make introductions 4. Create an Offer Letter
Materials
Draft an email to introduce your new Streamline the onboarding process by asking Found the perfect candidate? Use Copilot to
employee to their team. Copilot to generate learning paths according quickly draft an offer letter for you to send.
to individual roles, learning styles and career
goals

Copilot Copilot Copilot

Prompt: Draft an email to introduce Jane

Prompt: Develop a 3-month plan to onboard Prompt: Draft a customizable template for
to their new team based on this information
a new hire. an offer letter.
from their resume.

1
Access Copilot at copilot.microsoft.com or the Microsoft Copilot mobile app and set toggle to “Web”.
 Available Scenario
HR | Managing internal job transitions with:
Copilot for Microsoft 365 (with Copilot Studio level:
Extend
plug-ins)

Reduce onboarding Employee

KPIs impacted Cost per hire Improve eNPS Value benefit Cost Savings
time experience

3. Create onboarding
1. Quickly screen candidates 2. Schedule interviews
materials
Prompt Copilot1 to surface the most qualified Use Copilot1 to help schedule follow up Use Copilot in Word to create role-specific
candidates, referencing relevant documents interviews between candidates and hiring onboarding materials.
and pulling in employee information from the managers
HR system with Copilot Studio.

Copilot Copilot Copilot in Word

+Copilot Studio

Rapidly get up to speed on what Streamline scheduling multiple Reduce time to prepare, search and
summarize onboarding documents leading
candidates are most qualified for certain candidates for interviews by asking Outlook to increased employee satisfaction and helping
roles. to identify and schedule interviews. employees’ reach productivity quickly.

6. Prepare for performance

5. Collect feedback 4. Develop training material
reviews
Use Copilot in Word to generate reports for Analyze employee survey data with Copilot in Use Copilot in PowerPoint to generate
performance reviews. Excel and summarize key findings to refine learning paths according to individual roles,
delivery and relevance. learning styles and career goals.

Copilot in Word Copilot in Excel Copilot in PowerPoint

Focus on overall employee satisfaction Keep training relevant to the audience by

Compile data quickly and generate using Copilot to mine historical documents to
and onboarding experience by
individual review reports to support the update with relevant training content. As internal
responding to feedback and refining
process. systems change training can adapt to it more
onboarding and training. quickly.

1
Access Copilot at copilot.microsoft.com, the Microsoft Copilot mobile app, or the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
HR | Streamline benefits and compensation with:
Copilot for Microsoft 365 (with Copilot Studio level:
Extend
(Copilot Studio) plug-ins)

Employee
KPIs impacted Increase benefit usage Value benefit
experience

1. Conduct market research 2. Draft Executive Summary 3. Update financial model

Prompt Microsoft Copilot1 to research the Use Copilot in Word to draft a plan for talent Use Copilot in Excel to rapidly update the
latest industry trends and data regarding planning to share with your executive staff financial model.
competitive compensation ranges and for next fiscal year.
benefits.

Copilot Copilot in Word Copilot in Excel

+Copilot Studio

Turn a few thoughts on the data collected

Quickly generate tables of average Make rapid updates to the financial model
into
salaries by market for different roles. with Copilot.
a detailed planning document.

6. Communicate changes 5. Revise policy document 4. Present updates

Use Copilot in Outlook to draft an email Use Copilot in Word to revise relevant Use Copilot in PowerPoint to generate an
communicating changes to the compensation sections of the compensation policy executive presentation detailing the new
policy to department managers. document. compensation strategy.

Copilot in Outlook Copilot in Word Copilot in PowerPoint

Draft an email to a group of department Turn the bullet points from the executive Create a set of slides from the meeting
managers that provides detail and timelines
regarding employee annual reviews and changes
presentation into text for the policy notes to document the strategy and then
to the overall compensation and benefits. document. copy in the Excel charts.

1
Access Copilot at copilot.microsoft.com or the Microsoft Copilot mobile app and set toggle to “Web”.
 Available Scenario
HR | Improve organizational health metrics with:
Copilot for Microsoft 365 (with Copilot Studio level:
Extend
plug-ins)

Employee
KPIs impacted Improve eNPS Employee turnover rate Value benefit Cost Savings
experience

1. Identify data sources 2. Analyze data 3. Develop action plan

Identify multiple data sources which can provide Analyze data, summarize findings, provide Leverage insights and Copilot suggestions in
insights into manager capability, employee insights and create organizational health Word to draft a detailed, actionable
relations, and organizational health. Use Copilot1 reports. implementation plan to improve
to access HR system data using plugins built in organizational health.
Copilot Studio.

Copilot Copilot in Excel Copilot in Word

+Copilot Studio

Highlight org health scores yellow where YoY I am a human resources consultant. Create
Use Copilot to rapidly locate relevant data % change exceeded -20%. Generate an org an action plan based on the organizational
which indicates and impacts organization health report that summarizes the data and
health report and insights. Include tasks and
health. insights into 9 components: leadership,
innovation, etc. timelines.

5. Meet with Key

6. Communicate to managers 4. Generate the proposal
Stakeholders
Communicate plan and actions to Lead the meeting to present the plan, Generate a presentation detailing the plan
organization managers and monitor leveraging Copilot in Teams to suggest topics which includes images and tables.
engagement. and take notes.

Copilot in Outlook Copilot in Teams Copilot in PowerPoint

Use Copilot during the meeting to “list main Create a presentation from file
Draft with Copilot: an email to a group of
ideas we discussed” and then review [plan.docx] for HR leaders that clearly
managers that provides detail and timelines
the AI notes “Follow-up tasks” after the articulates how the plan will enable
regarding organizational health action plans.
meeting to finalize the plan. organizational health goals.

1
Access Copilot at copilot.microsoft.com, the Microsoft Copilot mobile app, or the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
HR | Candidate Search with: level:
Extend
Copilot for Microsoft 365 (with Copilot Studio
plug-ins)

Employee
KPIs impacted Cost per hire Value benefit Cost Savings
experience

3. Identify potential
1. Conduct market research 2. Create job description
candidates
Use Copilot1 to research the latest industry Ask Copilot1 to create a job description which Indicate keywords from your job description
trends and data regarding competitive job leverages the market research gathered. to surface potential candidates. Use Copilot2
descriptions. to access HR system data using plugins built
in Copilot Studio.

Copilot Copilot Copilot

+Copilot Studio

Generate a table of key skills and Generate a software engineer job

experience required for software engineers Rapidly identify possible candidates
description that includes skills,
at a large enterprise technology company based which meet role job requirements reducing
qualifications, and responsibilities. Reference
on market trends and leading technology time spent from hours to minutes.
companies. jobs at large, enterprise tech companies.​‌

5. Communicate potential
6. Finalize candidate slate 4. Create candidate slate
candidates
Engage with hiring manager to gain feedback Draft and send email to hiring manager Draft a detailed view of potential candidates
on candidate slate and align on next steps. sharing potential candidates and insights and insights so a hiring manager can easily
with them for consideration. evaluate candidates.

Copilot in Teams Copilot in Outlook Copilot in Word

“Recap the meeting” to review key topics “Draft with Copilot” an email to the hiring
Summarize these candidates in a table
and “List action items”. Paste actions in manager that summarizes the candidates
format with a pros column and a cons
the meeting chat and tag owners to confirm and the key insights for hiring consideration.
column.
or update the candidate slate. (Include the table from the Word document.)

1
Access Copilot at copilot.microsoft.com or the Microsoft Copilot mobile app and set toggle to “Web”.
2
Access Copilot at copilot.microsoft.com, the Microsoft Copilot mobile app, or the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
HR | Resolving employee issues with:
Copilot for Microsoft 365 level: Buy

Employee
KPIs impacted Improve eNPS Employee turnover rate Value benefit Cost Savings
experience

1. Summarize the challenge 2. Access resources 3. Brainstorm ideas

Aggregate a long and complex email thread Ask Copilot1 to help identify relevant and Create a list of questions needed to help
of conversations to create a holistic and timely resources to help solve for this solve this challenge.
summarized view of the challenge (employee common or complex challenge.
engagement, retention, diversity and
inclusion, etc.).

Copilot in Outlook Copilot Copilot in Loop

Use plugins built in Copilot Studio to rapidly Prior to a scheduled meeting, have
Select “Summarize” to understand key
identify relevant and timely data on how teammates build upon the Loop: brainstorm
issues and proposed solutions from lengthy
to approach common or complex employee ideas to address the HR situation and
email threads related to the HR challenge.
scenarios. challenge.

6. Communicate to 4. Collaborate with

5. Update documentation
teammates teammates
Communicate plan and actions to Update the approach, procedure and/or Schedule a meeting to discuss the options.
appropriate stakeholders and monitor policy documentation as applicable.
engagement.

Copilot in Outlook Copilot in Word Copilot in Teams

Draft with Copilot: an email to Leverage the meeting feedback to update Use Copilot during the meeting to “list main
stakeholders that provides detail and the approach. Rewrite this section to ideas we discussed” and then review
timelines regarding action plans to address update the “days to respond” from 5 the AI notes “Follow-up tasks” after the
the HR challenge. business days to 3 business days. meeting to confirm or update the approach
and documentation.

1
Access Copilot at copilot.microsoft.com, the Microsoft Copilot mobile app, or the Copilot app in Teams, and set toggle to “Web”.
 Available Scenario
HR | Update policy document with:
Copilot for Microsoft 365 level: Buy

Increase benefit Employee

KPIs impacted Improve eNPS Employee turnover rate Value benefit Cost Savings
usage experience

1. Identify changes in policy 2. Draft new FAQ 3. Create plan

Compare the new and old policy documents Create a new FAQ document that explains Ask Copilot1 to create a plan for change
and make a table of the key policy the changes and explores what impact it may management that updates curated content,
differences. Have Copilot1 translate the have on employees and what concerns they employee-facing material, and advisor
results into local language. might have. training and communications

Copilot Copilot in Word Copilot

Compare policies in these two Make a list of likely questions that Using the insights from the previous step, prompt
documents and make a table of the key employees will have about this change in Copilot: I am a human resources support advisor.
policy differences. Translate the results to policy and provide answers to each question Create a change management plan based on
Spanish. based on the policy document. the updated policy. Include tasks and timelines.

6. Continuous improvement 5. Collect feedback 4. Develop material

Communicate changes to Support Advisors Engage with support advisors to depict the Build training materials which are tailored to
and monitor impact/feedback.​ data, plan, actions and support material the case and employee/role.
required to implement the new policy.

Copilot in Outlook Copilot in Teams Copilot in PowerPoint

Draft with Copilot: a reply email to the Use Copilot during the meeting to “list main Create a presentation from file
department that provides detail and ideas we discussed” and then review the AI [plan.docx] and relevant documents about
notes “Follow-up tasks” after the meeting to
timelines regarding improvements and confirm or update the approach and
[insert case] to train Human Resources
changes. documentation. support advisors on applying the new policy.

1
Access Copilot at copilot.microsoft.com, the Microsoft Copilot mobile app, or the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
HR | Automate benefits query management with:
Copilot for Microsoft 365 and Copilot Studio level:
Extend

Onboarding Cost Employee

KPIs impacted Turnover rate Value benefit
time savings experience

1. Create a chatbot 2. Benefits analysis 3. Recommendation

Create a custom copilot accessing HR information Employees may ask the chat bot for more Employees can provide the chat bot with
in Microsoft Graph with guided prompts around details about specific benefits. desired outcomes and ask Copilot to pull
employee benefits. Once ready, launch it as an specific benefits recommendations.
app in Microsoft Teams for your organization.

Microsoft Teams Microsoft Teams Microsoft Teams

+Copilot Studio +Copilot Studio +Copilot Studio

Providing tailored recommendations

Streamline the benefits inquiry process Match employee criteria with the best
helps employees make informed decisions
with instant, 24/7 chatbot assistance. available benefits packages.
about their benefits.

6. Reporting 5. Administration assistance 4. Feedback collection

Use Copilot in Word to help draft a report on Leveraging the feedback collected, use Use the chat bot to automate the feedback
employee sentiment regarding resources, Copilot in Word to update employee collection in a database for analysis.
benefits, and potential improvements. resources and knowledge bases.

Copilot in Word Microsoft Teams

Copilot in Word +Copilot Studio

Reporting on trends and satisfaction

Streamlining the administration process Gathering feedback allows for the
helps HR to continually improve the benefits
reduces HR workload and improves accuracy. refinement of the recommendation process.
offerings.

1
Access Copilot at Copilot.Microsoft.com or from the Windows taskbar or Edge browser and set toggle to “Web”.
2
Access Copilot at Copilot.Microsoft.com, from the Windows taskbar or Edge browser, or in the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
a a a

A day in the life of a HR Manager with:

Copilot for Microsoft 365 (with Copilot Studio
plug-ins)
level:Extend

Benefits ~1 hour per week Areas of investment: Creative solutions Job satisfaction and output

8:00 am 9:35 am
10:00
am
Omar
leads HR for a
Omar starts the day at home with an At the office, Omar uses Copilot in Teams to Omar asks Copilot in Word to summarize the
interview for a new bank teller candidate. He summarize some chat threads that occurred organization’s new compliance handbook to regional bank
uses Copilot1 to research the candidate’s overnight and can quickly assess the ensure it has the key points. He then
previous companies. situation and provide guidance to his team to commands Copilot to fill in the missing
address the issue. sections.

Copilot Copilot in Teams Copilot in Word

What are some good follow up

Summarize this thread and include the Summarize the [Contoso Compliance
questions
key issues and suggestions for resolution Handbook] in about four paragraphs for an
to learn more about this person’s skills
along with who had the suggestions. executive and provide a list of key points.
and experience?

4:00 pm 2:00 pm 1:00 pm

Omar has missed a few calls and emails. Omar commands Copilot in PowerPoint to Omar prompts Copilot1 for details about open
He prompts Copilot1 to summarize recent add headcount, using a plugin built in Copilot
email threads and calls, and then uses a slide to his presentation that can be used Studio to pull in HR system data.
Copilot in Outlook to draft email responses. to explain the team’s initiatives.

Copilot in Copilot
Copilot Copilot in PowerPoint
Outlook +Copilot Studio

Add a column that averages the

Summarize this thread. Add a slide about potential HR initiatives.
other columns for each month.

1
Access Copilot at Copilot.Microsoft.com or from the Windows taskbar or Edge browser and set toggle to “Web”.
2
Access Copilot at Copilot.Microsoft.com, from the Windows taskbar or Edge browser, or in the Copilot app in Teams, and set toggle to “Work”.
 Available Scenario
a a a

A day in the life of a HR Manager with:

Copilot for Microsoft 365 (with Copilot Studio
plug-ins)
level:Extend

Benefits ~1 hour per week Areas of investment: Teamwork New initiatives

8:00 am 9:35 am
10:00
am
Aina
leads HR for a
Aina needs to create a job description that is Aina needs to review some applications she Aina decides to schedule an interview with a
a blend of two roles. has received. candidate. She uses Copilot to draft a reply technology
to their email that suggests times to meet.
company

Copilot in Word Copilot in Outlook Copilot in Outlook

Use the Schedule Interview prompt

Generate a job description that blends a
Use Summary by Copilot to pull out the suggestion to draft an email. Then ask
product marketing strategist and a sales role
key points in an email. Copilot to add some available times for an
based on these job descriptions.
interview.

3:00 pm 2:00 pm

Aina uses Copilot1 to help prepare for an Aina needs to schedule a vacation and uses
interview by generating interview questions Copilot1 to access a Copilot Studio plug-in to
based on the job description and resume. generate the request

Copilot
Copilot +Copilot Studio PTO system
plug-in

Generate 10 great interview questions I’d like to submit a vacation request for 3/24
based on [resume] and [job description] – 4/4

1
Access Copilot at Copilot.Microsoft.com, from the Windows taskbar or Edge browser, or in the Copilot app in Teams, and set toggle to “Work”.
Activity: Building query management system for HR using Gen AI
Instructions:
1.Group Formation:
a.Form groups of 3-4 members each, group will work together to brainstorm and
draft fictional HR policies for XYZ company .
2.Assigned Topics for drafting policies:
i. Dress Code and Appearance Policy
ii.Leave Policy
iii.Work from Home Policy
iv.Travel and Expense Reimbursement Policy
3.Content Creation:
a.Within their groups, participants should:
i. Draft policy statements, guidelines, and rules
ii.Use clear and concise language
4.PDF Creation:
a.Compile the policies into a well-organized PDF document.
5.Submission:
a.Send the created PDF file to [email protected]
Thank You