Unit Introduction
Welcome to the second unit of the Using AI in Make course!
In this unit, you’ll discover two Make AI apps that allow you to perform tasks using AI in your scenarios: Make AI Toolkit and Make AI Content Extractor.
You will learn: how Make AI Toolkit and Make AI Content Extractor work
how to use them in your scenarios
real-world applications of the Make AI apps
Make AI apps
Make AI apps are built-in tools within Make that use AI to process information and handle different tasks.
When using AI, you need to make sure to give precise instructions for what you
need. Make AI apps remove that step, handling the setup for you. So instead of figuring out instructions, you just input data and let the AI work its magic: summarize, extract, classify, translate, or perform other actions. This makes AI
easy to use in your scenarios, even if you have no experience working with it before.
Make AI Toolkit and Make AI Content Extractor
As you learned in the previous unit, there are two Make AI apps you can
include in your scenarios to perform various tasks.
You can use it to quickly handle text-based AI tasks like summarizing or splitting text into categories.
You can use it to extract text and key details (like names, IDs, totals or dates) from different types of files, such as documents, images or audio.
Why use Make AI apps
Here are some reasons why you should start using Make AI apps in your
scenarios:
1 Possibility to use Make’s AI Provider
2 Ready to use modules
Let’s check each of them in detail.
1 : PossibilitytouseMake’sAIProvider
Possibility to use Make’s AI Provider
In the previous unit, you learned that to use AI apps in Make, you must add a
connection to an AI provider by creating an account with an external service and
connecting it to Make. In the Make AI apps, you can select Make’s AI provider directly within the app, so you don’t need extra accounts or logins. Every subscription plan lets you use this
provider, giving you direct access to an LLM without relying on external services.
Ready to use modules
Normally, when you work with AI, you have to write a prompt that clearly tells the model what you want it to do. This can take some time, and if the instructions
are not precise, the AI might not give you the results you expect. With Make AI apps, you don’t need to worry about crafting prompts. Make created
various modules for the different actions that you can do with AI. Each module includes pre-built instructions for each task, which tell the AI what to do. You just add
your input, and the AI performs the action.
When using Make AI apps, only one module requires you to write a prompt: Make AI Toolkit> Simple Text Prompt. In this module you ask the AI something, so you must provide clear instructions. You will learn more about this module later.
Let’s learn more about each of Make’s AI apps.
Make AI Toolkit
Make AI Toolkit is an app that Make built to handle text-related AI tasks, which are operations where the AI processes a string of characters (text) as input. It could be anything from a short sentence to several paragraphs. Then, it returns text as output. Make AI Toolkit comes with pre-built prompts for the different actions, so you don’t
need to create instructions from scratch. You select a module, and the app processes your text to perform actions like analyzing sentiment, detecting a language, summarizing text, and more.
Using Make AI Toolkit
To use Make AI Toolkit, you need to make a connection to the AI provider.
This gives Make AI Toolkit access to an LLM, which allows it to process the text and give you a result. You can choose between two types of AI provider connections, you already learned this in the previous unit, so let’s recap.
Make’s AI Provider
You can use Make’s AI provider to connect to different models. This means Make will handle the connection for you. It is available for all plans, including Free.
External AI providers
If you are subscribed to a third-party AI provider, you can bring your own connection. It is available on the Pro plan and above.
Credit usage in your scenario is different depending on your AI provider connection. Here you will find more information on how the AI provider connection works and how credits are used for each option. After choosing Make’s AI Provider, you need to give it access to the LLM so it can process and understand the text, you need to decide which model (LLM) to choose. If you need a recap, you can go to the Mastering LLMs course.
There are three different models available when using Make’s AI Provider: Small, Medium, and Large. You can select the model depending on the type of task you
need to perform. When choosing the model, think about the complexity and length of the text, as well as the cost from processing the data.
Best practices If you are not sure about which model would give you the best results, you can start with the Small model to save credits. If it doesn’t give good results, work your way up until you get the outcome you are looking for. This way, you will not consume more credits than you need.
Here you will find more information on how this works:
CREDITS AND TOKENS
2.3.2: What can you do with Make AI Toolkit? You can select between nine different modules, depending on what you need to use AI for. Each module is set to perform a different action. Here is a list of each module along with a real-life example.
Simple Text Prompt
Sends a request for AI to answer, the AI interprets your request and returns a text response. This is the only module for which you provide the prompt. This gives you the freedom to ask the AI anything and perform multiple tasks.
Real-life example: Ask the AI to draft a professional email to a client confirming a meeting.
Extract information from text
Extracts specific pieces of data from a larger body of text. Real-life example: Extracts phone numbers and email addresses from a list of customer inquiries and saves them in a spreadsheet so they are easier to track and use.
Categorize text
Analyzes the content of your text and automatically assigns it to one or more categories you define. You set the list of categories, and the AI determines which one best fits each piece of text.
Real-life example: Automatically organizes incoming support tickets into three categories: Billing, Technical Issue, or General Inquiry so different customer service agents can handle each category.
Translate text
Converts text from one language to another. Real-life example: Translate a product manual from English into German.
Identify language
Detects the language the text is written in.
Real-life example: Determine the language of a customer request so that you can assign the ticket to the relevant support agent.
Summarize text
Takes long pieces of text and condenses them into a shorter, more digestible summary. It identifies the key points and main ideas, so you can quickly understand the content without reading everything in detail.
Real-life example: Create a one-paragraph summary of a 10-page research report so you can use it directly as the document’s summary. Analyze sentiment
Evaluates the tone of a text and classifies it as positive, neutral, or negative. It can help you identify trends in feedback or analyze customer opinions.
Real-life example: Analyze customer reviews of your product to see if people feel satisfied, frustrated, or indifferent.
Standardize text
Makes sure all your text follows the same style and format. It can adjust capitalization, spelling, tone, or date formats so your content looks uniform.
Real-life example: Convert all product descriptions to use the same date format, capitalization, and tone. Chunk text
Splits text into smaller, manageable pieces. It allows you to handle long documents efficiently, ensures accurate processing, and makes analysis or summarization more manageable.
Real-life example: Break down a long transcript into 500 word sections for easier analysis and summarization.
Example use cases
Now that you know how it works, you will see three scenarios. Each uses
modules from Make AI Toolkit to perform different tasks.
This is what the use cases will do:
1 Follow up and translate support ticket
2 Analyze sentiment
3 Classify emails
Let’s check each of them in detail. 1 : Followupandtranslatesupportticke t
Follow up and translate support tickets
Carla, the Customer Service Manager at Wonderwall Enterprises, collects customer feedback, questions, and requests through Typeform, an online form tool.
She creates a scenario that uses AI to check each response to see if it needs a follow-up and translates the message into English so her team can understand it.
Then, it creates a ticket in Zendesk, the ticket management system they use, to ensure the team tracks and addresses every case.
Typeform> Watch responses:
Starts the scenario when a user sends a new form.
Make AI Toolkit> Simple Text Prompt:
Asks the AI to determine if the customer is waiting for a response.
Filter:
Allows only responses that require a follow-up to pass to the next module.
Make AI Toolkit> Translate text:
Translates the responses that require a follow-up to English.
Zendesk> Create a ticket:
Using the translated text, it creates a ticket so the team can follow up on the response.
2: Analyze sentiment
Analyze sentiment Carla wants to handle emails from unsatisfied customers more efficiently by
connecting them directly to a customer service agent.
She creates a scenario using Make AI Toolkit to analyze the sentiment of each
email received. If the sentiment is negative, the scenario sends a notification to a customer service agent right away via Teams, the team’s communication platform.
Gmail> Watch Emails:
Checks for new emails.
Make AI Toolkit> Analyze sentiment:
Evaluates the tone of emails and determines if it’s positive, neutral, or negative.
Filter:
Allows only negative sentiment emails to pass to the next module.
Teams> Send a message:
Sends a notification to the customer service channel.
3: Classify emails
Classify emails Carla often receives emails that are meant for different departments.
To route them correctly, she builds a scenario that uses AI to classify each email
as an issue, inquiry, or feedback. The scenario then notifies the right team: Sales
handles inquiries, Marketing handles feedback, and Customer Support handles
issues.
Gmail> Watch Emails:
Checks for new emails.
Make AI Toolkit> Categorize text:
Analyzes the email text and assigns it to the categories: issue, inquiry, or feedback.
Router:
Allows you to branch the scenario into different paths.
Filters:
Only allows emails from certain categories to pass to the different modules.
Teams> Send a Message:
Sends a message to a team channel or a chat.
Make AI Content Extractor
Make AI Content Extractor is an app built by Make that processes non-text files,
such as documents, images, or speech. Each module has a pre-built instruction
for each task and returns text as the output.
The app can identify the most relevant details, such as names or dates, detect
objects and text in images, and even generate captions, descriptions, or tags. Make AI Content Extractor can interpret and organize the content, so you get
ready-to-use text. You can then pass these results into other modules inside your Make scenarios to automate further tasks. As you’ve learned, you can use Make AI Content Extractor to work with different file types.
Extracts text from PDFs, Word files, or scanned pages.
Analyzes images to detect objects, reads visible text, and describes the content. Converts audio recordings into text and translates it.
Earlier in the unit, you learned about the Make AI Toolkit. You might be wondering, how is this app different from Make AI Content Extractor? Make AI Toolkit works only with text. It analyzes the text you give it, understands
the context, and extracts or transforms the information you need.
On the other hand, Make AI Content Extractor can work with non-text files like PDFs, images, spreadsheets, or HTML. Its job is to extract the information inside
those files.
Now let’s discover everything you can do with Make AI Content Extractor.
Extractor?
What can you do with Make AI Content Extractor?
Make AI Content Extractor has 11 different modules: Here is a list of each module you can use with Make AI Content Extractor, along with a real-life example for each one.
Extract text from a document
Extracts text from PDFs, Word files, or other document types.
Real-life example: Extract the full text from a legal contract PDF so you can store it in a database for easy searching.
Extract information from an invoice
Identifies and extracts key invoice details.
Real-life example: Extract vendor name, invoice ID, and total amount from PDF invoices received from multiple suppliers, and send this information directly to the company’s payment system.
Extract information from a receipt Reads and extracts details like merchant name, date, and transaction amount from receipts.
Real-life example: Automatically extracts details from employee expense receipts so they can be logged directly into the expense management system.
Generate a caption for an image
Creates a concise, one-sentence caption describing the content of an image.
Real-life example: Generate captions for product photos to populate an e- commerce website to keep consistency in product description.
Generate captions for an image (advanced)
Produces up to 10 captions describing different parts or aspects of an image. Real-life example: Provide multiple caption options for marketing teams to choose from when posting product images on social media so that they can pick the one that fits best.
Describe an image
Provides a detailed description of an image.
Real-life example: Create detailed descriptions of images to improve accessibility in a website.
Extract text from an image
Reads printed or handwritten text from images and converts it into text. Real-life example: Converts notes into searchable text, making it easy to find specific information later.
Generate image tags
Creates a list of relevant keywords describing the content of an image.
Real-life example: Provides keywords for product images to improve searchability in an online store.
Detect objects in an image
Identifies objects in an image and marks their approximate location in the picture.
Real-life example: Detect cars in photos submitted by customers to verify that a vehicle is present before processing insurance claims. Transcribe an audio file
Converts an audio file into written text.
Real-life example: Transcribe customer service calls into text to search the text for recurring issues later.
Translate an audio file
Converts an audio file into English.
Real-life example: Translate a Spanish-language voice memo into English so an international team can review the conversation and respond appropriately.
Example use cases
It’s time to explore some practical use cases that include Make AI Content Extractor. Check them out, this will give you a better idea of what you can do with the different modules available. This is what the use cases will do:
1 Extract text from a document
2 Transcribe an audio file
3 Extract text from an image
Let’s explore each use case.
Extract text from a document Hans is a Sales Representative at Yellow Submakerine Inc., a software company. He stores client PDF contracts in HubSpot, a customer relationship management
(CRM) platform he uses to manage deals and client information. To save time, he creates a scenario that uses AI to pull the text from each PDF and identify the client name, contract ID, and renewal date from each contract. The scenario then updates
the deal record in HubSpot with those details, keeping everything organized and
easy to track.
HubSpot CRM> Watch Files:
Checks for new files.
Make AI Content Extractor> Extract text from a document:
Extracts text from the PDF contract so you can store the information in Hubspot.
Make AI Toolkit> Extract information from a text:
Uses the extracted text to identify the key details from the text: client name, contract ID and renewal date.
HubSpot CRM> Update a Deal:
Updates the details of the deal with the information extracted.
Transcribe an audio file Hans often receives requests from customers and records calls to keep a record
of approvals. He stores all audio files in Box, a cloud-based storage service. Hans created a scenario that uses Make AI Content Extractor to transcribe the call, and then saves the transcription to the customer’s case in Salesforce, a CRM, so he can
access and search it easily.
Make AI Content Extractor> Transcribe an audio file:
Converts the audio from the previous step into text.
Salesforce> Update a Record
Stores the transcription in the customer’s case for storage and access.
Box> Download a File:
Retrieves files from Box for processing.
Extract text from an image When customers encounter software bugs or need tech support, they often
send screenshots via email. Hans wants to capture all relevant information from these screenshots, so he creates a scenario that checks new email attachments, uses AI to extract text from the screenshots and creates a ticket, allowing the team to search and track tickets with the same error messages and answer efficiently.
Gmail> Watch Emails:
Checks for new emails.
Gmail> Iterate Attachments:
Retrieves each email attachment so the next module can process it individually.
Make AI Content Extractor> Extract text from an image:
Retrieves printed text from the attachments retrieved from the previous module.
Zendesk> Create a Ticket:
Generates a ticket containing the text from the previous module, this ensures the team addresses customer issues.
Continuetothewrapupforthisunit
Wrap up
Make AI apps integrate AI tasks directly into your scenarios. They provide ready-to-use modules with pre-built instructions, so you don’t need to craft detailed prompts. You can add a connection to Make’s AI Provider, without having to create or pay for any external accounts.
Make AI Toolkit lets you perform common text-based AI tasks in your scenarios without writing prompts from scratch. Its pre-built modules can analyze sentiment, detect language, summarize text, extract information, translate, and more, so you can quickly process and organize text, emails, or documents. You can use them in your scenarios to automate tasks like following up on support tickets, analyzing customer sentiment, or classifying emails.
Make AI Content Extractor processes the content of documents, images, and audio, and provides a text result. It works with non-text files, unlike Make AI Toolkit which works only with text. You can use Make AI Content Extractor in your scenarios to automate tasks like extracting contract details, transcribing calls, or reading error messages from screenshots, saving time and reducing errors.
Unit complete!
Nice! You have completed the second unit of the course. By now you should have an understanding of:
how Make AI Toolkit and Make AI Content Extractor work
how to use them in your scenarios
real-world applications of the Make AI apps
Next, you will apply everything you have learned and use AI apps in
your scenarios. Mark this task complete to continue to the next unit.