app.py

import openai
import random
import gradio as gr
import os

openai.api_key = os.environ.get['OPENAI_API_KEY']

manual = r"""Input your variables in the fields on the left. The last two (Extra input and Team) are optional, but recommended. You can use it to steer the prompt into a certain direction by adding more requirements.
"""

# Define the messages for step_1
SystemPrompt_10 = r""""You are a script writer and prompt engineer creating engaging YouTube tutorial scripts on how to use ChatGPT for specific tasks in a corporate environment. 

Given some instructions, a hook, a title, and an introduction, you design the execution part of the script. Come up with 2 learning objectives about tasks that can be executed with the help of ChatGPT. This is the structure you'll follow:

- Learning Objectives Summary
- Learning Objective 1
- Learning Objective 2
- Learning Objective 3 (this third one is optional)
- Conclusion

The learning objectives state the specific learning objectives that will be covered in the tutorial. They must always start with an Action Verb that specifies what the learner should do. Common verbs include "demonstrate," "explain," "analyze," "solve," "create," "compare," and so on. The choice of verb reflects the type of cognitive or skill-based activity the learner should engage in.

Make sure to also include:
- the exact prompt(s) the user needs to type into ChatGPT. These prompts need to be automatic prompts and follow best practices for prompt engineering like "define a role (you are an expert marketer, you are an expert recruiter) " and "provide necessary context." The first prompt always needs to be an automatic prompt, and end with "Ask me as many questions as you need until you have enough information to provide me with the desired output.";
- realistic sample data for a case study, if applicable;
- for every learning objective, explanations of concepts that the audience may not understand. """

UserPrompt_10 = r""""Task: Crafting product update checklists on product updates and new features for customer onboarding
Task Description: Review the latest product updates and new features, ensuring a comprehensive understanding of their functionalities and benefits.
Create detailed checklists that outline step-by-step procedures for clients to effectively integrate and utilize these updates and features in their existing systems or processes.
Ensure the checklists are clear, user-friendly, and tailored to support customers of varying technical proficiency, facilitating smooth onboarding and optimal use of the new product capabilities.
Additional instructions: The tutorial is aimed at employees in a Client Support department."

HOOK:
If you're in client support, you know how tricky it can be to explain new product features to a wide range of customers. Imagine having a tool that helps you create easy-to-follow checklists for every update, making the customer onboarding process a breeze. With ChatGPT, creating user-friendly guides for product updates just got a lot easier!

TITLE:
Create Easy-to-Follow Product Update Checklists with ChatGPT: A Guide for Client Support Teams

INTRO:
Hello there, Client Support team members! It's a known fact that when a new product update rolls out, it's your responsibility to ensure that customers understand and can effectively use these new features. This isn't simply about telling them what's new; it's about guiding them through the process, step by step, ensuring they can integrate these updates into their systems or processes, regardless of their technical proficiency.

But how do you create clear, user-friendly checklists that cater to customers with varying levels of expertise? That's where ChatGPT comes in. In this tutorial, we'll walk you through how to use ChatGPT to develop comprehensive, easy-to-understand checklists for every product update. We'll show you how this tool can help you break down complex updates into simple, manageable steps for your customers. Are you ready to revolutionize your customer onboarding process and make product updates a seamless experience for your clients? Let's get started!"""

AssistantPrompt_10 = r""""*LEARNING OBJECTIVES SUMMARY*
You will learn how to:
-	Create an ideal job description
-	Extract SEO-friendly keywords 
-	Craft a job ad 

Let's dive right in!

*LEARNING OBJECTIVE 1*
First, let's create the ideal job description. You can't create the perfect job ad without first creating the job description. 

A job description is a document that outlines the key responsibilities, qualifications, and skills required for a specific role, along with details about working conditions and reporting relationships. 

Creating a comprehensive job description is critical for attracting the right candidates, setting expectations, and ensuring both the employee and employer understand the role's responsibilities.   

To create a basic job description, you at least need to define the job title, the overview of the role, and the role's key responsibilities.

 It's also preferable to already identify any other details about the role, like qualifications and skills, experience level, salary and benefits, reporting structure, and working hours.    

Let's say you've been tasked to recruit a 'Digital Marketing Manager'.

Gather as much information as you can on the specifics of this job from the person that made the request. 

For this tutorial, we’ll be using the following sample information. 

The job title is "Digital marketing manager."

Here's an overview of the role: "The digital marketing manager is responsible for managing the overall digital marketing strategy of xpedite.ai. He/she is also responsible for:
-	overseeing online marketing campaigns and other digital projects to optimize online brand presence and boost revenue.
-	managing a team of 3 digital marketers." 

The key responsibilities of the role are "propose and execute digital marketing strategies, manage and oversee digital channels, monitor and measure ROI and KPIs of online campaigns."

Now, we're ready to write our prompt. 

Type: 
“You are an expert recruiter. Your current task is to create a job description for a digital marketing manager. Here are some additional details to consider when writing the job description: (Then, paste the details you gathered previously.) 

Ask me as many questions as you need until you have enough information to provide me with the desired output."

Then press the send button.


ChatGPT asked relevant questions until it had enough information to come up with the desired output.

Make sure to review and revise this job description so it effectively communicates your expectations. 

*LEARNING OBJECTIVE 2*
Now, we're ready to extract SEO-friendly keywords. With this description, we can proceed to extract keywords which potential candidates might use when searching for this job.

Keywords are important because they make your job ad discoverable. The best keywords should resonate with the job seekers' search behavior.

Let's get keywords from ChatGPT. 

Type:
"Provide keywords related to the role of Digital Marketing Manager based on the description you've provided me."

Then, press the send message button. 

ChatGPT came up with a couple of keywords that we can use when posting our final job ad. 

We can also incorporate these keywords into the job ad itself.

*LEARNING OBJECTIVE 3*
Finally, let's craft a job ad. With our description and keywords in hand, let's craft a job ad that integrates them both seamlessly.  

Type:
"Using the job description and the keywords you've provided, draft a job ad for a Digital Marketing Manager."

 Then, press the send button.   

Here's the response. 


ChatGPT drafted a job ad that contains specific details about the job description and incorporates keywords so that it's easier to find on the internet. 

Make sure to review and revise the ad and verify that all information is accurate before posting it online. 

*CONCLUSION*
That's it!

By leveraging the power of ChatGPT-4, HR professionals can create job advertisements that stand out and attract top talent.

Remember, in the digital age of recruitment, your job ad is your first impression. So make it count!"""

UserPrompt_11 = r""""Task: Crafting product update checklists on product updates and new features for customer onboarding
Task Description: Review the latest product updates and new features, ensuring a comprehensive understanding of their functionalities and benefits.
Create detailed checklists that outline step-by-step procedures for clients to effectively integrate and utilize these updates and features in their existing systems or processes.
Ensure the checklists are clear, user-friendly, and tailored to support customers of varying technical proficiency, facilitating smooth onboarding and optimal use of the new product capabilities.
Additional instructions: The tutorial is aimed at employees in a Client Support department."

HOOK:
If you're in client support, you know how tricky it can be to explain new product features to a wide range of customers. Imagine having a tool that helps you create easy-to-follow checklists for every update, making the customer onboarding process a breeze. With ChatGPT, creating user-friendly guides for product updates just got a lot easier!

TITLE:
Create Easy-to-Follow Product Update Checklists with ChatGPT: A Guide for Client Support Teams

INTRO:
Hello there, Client Support team members! It's a known fact that when a new product update rolls out, it's your responsibility to ensure that customers understand and can effectively use these new features. This isn't simply about telling them what's new; it's about guiding them through the process, step by step, ensuring they can integrate these updates into their systems or processes, regardless of their technical proficiency.

But how do you create clear, user-friendly checklists that cater to customers with varying levels of expertise? That's where ChatGPT comes in. In this tutorial, we'll walk you through how to use ChatGPT to develop comprehensive, easy-to-understand checklists for every product update. We'll show you how this tool can help you break down complex updates into simple, manageable steps for your customers. Are you ready to revolutionize your customer onboarding process and make product updates a seamless experience for your clients? Let's get started!
"""

AssistantPrompt_11 = r""""*LEARNING OBJECTIVES SUMMARY*
By the end of this tutorial, you will be able to:

1. Use ChatGPT to interpret product updates and new features
2. Generate detailed, user-friendly checklists for product onboarding

*LEARNING OBJECTIVE 1*
Firstly, let's get a comprehensive understanding of product updates and their benefits. 

Product updates can range from minor tweaks to significant overhauls. It's crucial to understand these updates thoroughly before creating a checklist to guide customers.

Suppose we have a recent update for an email marketing software called 'MailBlitz'. The update includes a new feature that allows users to segment their email lists based on customer behavior and an improved user interface.

To get a comprehensive understanding, feed this information to ChatGPT.

Type into the ChatGPT: 
"You are a software expert. I need a detailed explanation of the latest updates on an email marketing software called 'MailBlitz'. The updates include a new feature allowing users to segment their email lists based on customer behavior and an improved user interface. Ask me as many questions as you need until you have enough information to provide me with the desired output."

Then, press the send button. 

ChatGPT will ask relevant questions until it has enough information to provide a detailed explanation.

*LEARNING OBJECTIVE 2*
Next, let's move on to create a user-friendly checklist for these updates. This checklist is meant to guide users through the process of effectively integrating and using these updates, regardless of their technical proficiency.

Type into the ChatGPT: 
"Now that we understand the update, let's create a step-by-step checklist to help users integrate these updates into their systems. The checklist should be clear, user-friendly, and cater to users with varying levels of technical proficiency."

Then, press the send button. 

ChatGPT will generate a comprehensive checklist, breaking down complex procedures into simple, digestible steps.

*CONCLUSION*
And there you have it! With the help of ChatGPT, you've just created a detailed, user-friendly checklist to guide your customers through the latest product updates. Remember, the key is to understand the updates thoroughly before creating the checklist. Use this tool to your advantage and make onboarding a seamless experience for your clients. Happy onboarding!"""


# Define messages for stepTwo
SystemPrompt_20 = r"""You are a script writer and prompt engineer creating engaging YouTube tutorial scripts on how to use ChatGPT for specific tasks in a corporate environment. 

Given some instructions, a hook, a title, and an introduction, you design the execution part of the script. Come up with 2 learning objectives about tasks that can be executed with the help of ChatGPT. This is the structure you'll follow:

- Learning Objectives Summary
- Learning Objective 1
- Learning Objective 2
- Learning Objective 3 (this third one is optional)
- Conclusion

The learning objectives state the specific learning objectives that will be covered in the tutorial. They must always start with an Action Verb that specifies what the learner should do. Common verbs include "demonstrate," "explain," "analyze," "solve," "create," "compare," and so on. The choice of verb reflects the type of cognitive or skill-based activity the learner should engage in.

Make sure to also include:
- the exact prompt(s) the user needs to type into ChatGPT;
- realistic sample data for a case study, if applicable."""

UserPrompt_20 = r"""Topic of the tutorial: How to create a Job Ad with ChatGPT.
Additional instructions: Use digital marketing manager as an example role.

*HOOK*
If you want your job ads to be found by the right candidates, you have to incorporate the right SEO-friendly keywords.

But how do you accomplish this without a lot of experience in Search Engine Optimization? 

Well, with ChatGPT. 

*TITLE*
Create a Job Ad with ChatGPT

*INTRO*
In this digital age, getting your job ad to stand out amongst the rest is crucial to finding the best candidates.

Today, we'll show you how to use ChatGPT to create a job ad that not only resonates with your ideal candidates, but also ranks high with SEO-friendly keywords."""

AssistantPrompt_20 = r"""*LEARNING OBJECTIVES SUMMARY*
You will learn how to:
-	Create an ideal job description
-	Extract SEO-friendly keywords 
-	Craft a job ad 

Let's dive right in!

*LEARNING OBJECTIVE 1*
You can't create the perfect job ad without first creating the job description. 

Creating a comprehensive job description is critical for attracting the right candidates, setting expectations, and ensuring both the employee and employer understand the role's responsibilities.   

To create a basic job description, you at least need to define the job title, the overview of the role, and the role's key responsibilities.

 It's also preferable to already identify any other details about the role, like qualifications and skills, experience level, salary and benefits, reporting structure, and working hours.    

Let's say you've been tasked to recruit a 'Digital Marketing Manager'.

Gather as much information as you can on the specifics of this job from the person that made the request. 

For this tutorial, we’ll be using the following sample information. 

The job title is "Digital marketing manager."

Here's an overview of the role: "The digital marketing manager is responsible for managing the overall digital marketing strategy of xpedite.ai. He/she is also responsible for:
-	overseeing online marketing campaigns and other digital projects to optimize online brand presence and boost revenue.
-	managing a team of 3 digital marketers." 

The key responsibilities of the role are "propose and execute digital marketing strategies, manage and oversee digital channels, monitor and measure ROI and KPIs of online campaigns."

Now, we're ready to write our prompt. 

Type: 
“You are a recruiter. Create a job description for a digital marketing manager.

Here are some additional details to consider when writing the job description:”

Then, paste the details you gathered previously.

Then press the send button.


ChatGPT wrote some basic details, the overview of the role, and key responsibilities of the role. 

It also included [basic and preferred qualifications], [what's in it for the applicant], and [more instructions].

Make sure to review and revise this job description so it effectively communicates your expectations. 

*LEARNING OBJECTIVE 2*
With this description, we can proceed to extract keywords which potential candidates might use when searching for this job.

Keywords are important because they make your job ad discoverable. The best keywords should resonate with the job seekers' search behavior.

Let's get keywords from ChatGPT. 

Type:
"Provide keywords related to the role of Digital Marketing Manager based on the description you've provided me."

Then, press the send message button. 

ChatGPT came up with a couple of keywords that we can use when posting our final job ad. 

We can also incorporate these keywords into the job ad itself.

*LEARNING OBJECTIVE 3*
With our description and keywords in hand, let's craft a job ad that integrates them both seamlessly.  

Type:
"Using the job description and the keywords you've provided, draft a job ad for a Digital Marketing Manager."

 Then, press the send button.   

Here's the response. 


ChatGPT drafted a job ad that contains specific details about the job description and incorporates keywords so that it's easier to find on the internet. 

Make sure to review and revise the ad and verify that all information is accurate before posting it online. 

*CONCLUSION*
That's it!

By leveraging the power of ChatGPT-4, HR professionals can create job advertisements that stand out and attract top talent.

Remember, in the digital age of recruitment, your job ad is your first impression. So make it count!"""


# Function to make API call
def api_call(messages, temperature=0.9, model="gpt-4-1106-preview"):
    return openai.ChatCompletion.create(
        messages=messages,
        temperature=temperature,
        model=model
    ).choices[0].message.content


# Function to be called by Gradio interface
def process_inputs(EPA_title, Task_Description, Extra_input, Department):
    # Check if EPA_title is empty
    if not EPA_title:
        return manual
    if not Extra_input:
        stepOne = [
            {"role": "system", "content": SystemPrompt_10},
            {"role": "user", "content": UserPrompt_10},
            {"role": "assistant", "content": AssistantPrompt_10},
            {"role": "user", "content": UserPrompt_11},
            {"role": "assistant", "content": AssistantPrompt_11},
            {"role": "user", "content": f"""Task: {EPA_title}.
             Task Description: {Task_Description} 
             Additional Instructions: The tutorial should be aimed at employees in a {Department} department."""}
        ]
        Script_1st_half = api_call(stepOne, 0.7)

        stepTwo = [
            {"role": "system", "content": SystemPrompt_20},
            {"role": "user", "content": UserPrompt_20},
            {"role": "assistant", "content": AssistantPrompt_20},
            {"role": "user", "content": f"""Task: {EPA_title}.
             Task Description: {Task_Description} 
             Additional Instructions: The tutorial should be aimed at employees in a {Department} department."""}
        ]
        Script_2nd_half = api_call(stepTwo, 0.7)
        Script_complete = Script_1st_half + "\n\n" + Script_2nd_half
        return Script_complete

    else:
        # Step 1: User input and first API call ~5secs
        stepOne = [
            {"role": "system", "content": SystemPrompt_10},
            {"role": "user", "content": UserPrompt_10},
            {"role": "assistant", "content": AssistantPrompt_10},
            {"role": "user", "content": UserPrompt_11},
            {"role": "assistant", "content": AssistantPrompt_11},
            {"role": "user", "content": f"""Topic of the tutorial: {EPA_title} 
            Additional instructions: {Extra_input}
            The tutorial is aimed at employees in a {Department} department."""}
        ]
        Script_1st_half = api_call(stepOne, 0.7)

        stepTwo = [
            {"role": "system", "content": SystemPrompt_20},
            {"role": "user", "content": UserPrompt_20},
            {"role": "assistant", "content": AssistantPrompt_20},
            {"role": "user", "content": f"""Topic of the tutorial: {EPA_title}
            Additional instructions: {Extra_input}
            The tutorial is aimed at employees in a {Department} department.

            {Script_1st_half}"""}
        ]
        Script_2nd_half = api_call(stepTwo, 0.7)
        Script_complete = Script_1st_half + "\n\n" + Script_2nd_half
        return Script_complete

departments = {
    "3D": "3D Modeling",  # Label and value differ
    "Client Support": "Client Support",
    "Data Entry & Back Office": "Data Entry & Back Office",
    "Education Administration (ENI)": "Education Administration",  # Label and value differ
    "Finance": "Finance",
    "Marketing": "Marketing",
    "HR": "HR",
    "IT": "IT",
    "Legal": "Legal",
    "Product Management": "Product Management",
    "Project Management/Coordination": "Project Management/Coordination",
    "Process Management": "Process Management",
    "QA": "QA",
    "R&D": "R&D",
    "Sales": "Sales",
    "Software": "Software",
    "Video Production": "Video Production",
    "Visual Management": "Visual Management"
}

# Create the Gradio interface
iface = gr.Interface(
    fn=process_inputs,
    inputs=[
        gr.Textbox(lines=2, label='EPA title Here... (use the phrasing: "How to do X using ChatGPT")'),
        gr.Textbox(lines=2, label='Task Description'),
        gr.Textbox(lines=2, label='Extra input Here... (optional, put any extra requirements or relevant context)'),
        gr.Dropdown(choices=departments, label="Select Team", searchable=True)
    ],
    outputs=gr.Textbox(label="Script", show_copy_button=True)
)

iface.launch(share=True)
iface.launch()