An AI Agent is a customizable, automated system created in MindStudio that performs specific tasks using artificial intelligence. Think of it as your digital assistant, capable of handling repetitive, data-driven, or creative tasks to save you time and effort.

Key Features of an AI Agent

Task Automation

AI Agents are designed to execute tasks automatically based on your configurations. Most of the tasks that are executed leverage AI models to complete the task.

AI-Powered Actions

AI Agents utilize AI models to perform actions such as text generation, summarization, image generation, data extraction, and analysis, and more.

Customizable Workflows

Each AI Agent is built using workflows that define its behavior. Workflows are a sequence of tasks that the Agent executes each time it runs.

Dynamic and Flexible

AI Agents leverage variables, triggers, and conditions to customize the Agent's output based on specific inputs or contexts.

Integration-Ready

AI Agents can be integrated with other platforms and tools via APIs or third-party services like Zapier.

How Does an AI Agent Function?

1. Inputs: AI Agents receive inputs like user provided data, uploaded files, or launch variables.

2. Workflow Execution: The Agent follows a defined sequence of actions to process the inputs.

   Example: Search for the latest tech news → Summarize the results → Send an email.

3. Outputs: The AI Agent ends its workflow with a resulting response based on the workflow execution.

Why Use AI Agents?

• Efficiency: Automate time-consuming tasks and reduce manual effort.

• Scalability: Handle large volumes of data or repetitive tasks seamlessly.

• Customization: Tailor workflows to your specific needs with ease.

• Integration: Connect AI Agents to other systems for end-to-end automation.

An AI Agent is your go-to solution for leveraging AI to simplify complex tasks, boost productivity, and drive innovation in your workflows.

Example Use Cases with workflows

1. Content Summarization

   • AI Agent: Summarizes long articles into concise bullet points.

   • Workflow: Fetch content → Analyze content→ Summarize → Deliver via email.

2. Data Processing

   • AI Agent: Processes survey data to highlight key insights.

   • Workflow: Import data → Analyze trends → Generate report.

3. Customer Support

   • AI Agent: Automates FAQ responses for a website.

   • Workflow: Receive question → Match with answers → Respond.

More Use Case Examples

AI Agents have countless use cases. In the next article, we’ll be showcasing an extensive list of uses cases that span across several job functions and tasks.

MindStudio's intelligence infrastructure gives everyone—from developers to non-technical teams—the power to build and deploy specialized AI Agents.

Get Started

If you are new to MindStudio, start here to learn the essentials and make your first AI Agent.

For Developers

MindStudio has best-in-class developer tools to build scalable AI Agents for a variety of fuzzy computing tasks.

Capabilities

MindStudio offers powerful capabilities that enable you to create, test, and deploy sophisticated AI Agents with ease.

Additional Resources

Find helpful resources to enhance your experience with MindStudio.

The Workflow Generator feature in MindStudio is a powerful tool that simplifies the creation of AI Agents. By describing your desired outcome, the feature automatically generates a detailed workflow plan, complete with prompts, variables, automation steps, and custom functions.

How to use the Workflow Generator

Step 1: Access the Workflow Generator

1. Open a new or existing project in MindStudio.

2. Navigate to the bottom bar and select the Generate Workflow button.

Step 2: Describe your AI Agent

A modal will appear asking you to describe what you want your AI Agent to do. Click Generate to let MindStudio process your request.

Example:

Ask the user for their name and birthday, then write them a horoscope.

Step 3: View the AI Generated Workflow Proposal

Once the workflow is generated:

Left Panel: Displays a detailed plan of the workflow, including the system prompt, steps, and variable definitions.

Right Panel: Translates the plan into executable automation steps, complete with code and configurations.

Key components generated by AI:

• System Prompt: Pre-configured text to guide the workflow.

• Variables: Input fields like name and birthdate.

• Custom Functions: Auto-generated JavaScript or Python functions, such as one to compute a zodiac sign from a birthdate.

• Automation Steps: A preview of the blocks added to the workflow.

Step 4: Approve Proposal and Test

1. If the generated workflow looks good, click Accept and Build.

2. MindStudio will automatically construct the workflow on the canvas.

Step 5: Debug and Refine

• Use the Debugger to analyze the execution of the workflow:

  • Review logs for actions, variable updates, and custom function outputs.

  • Identify and fix any issues or edge cases.

• Refine the workflow by adjusting blocks, variables, or custom function code.

• Use the preview mode to test the draft:

  • Input test data.

  • Observe the execution in real-time, including the generated outputs.

Important: Review Your AI Generated Builds

While the Generate Workflow feature in MindStudio is a powerful tool for automating workflow creation, it’s essential to review all generated outputs when accepting an auto-generated build. The feature uses AI to interpret your descriptions and generate workflows, but it is still prone to occasional errors or misconfigurations.

Potential Issues to Watch For

Incorrect or Missing Variable Connections

Variables used in prompts, automation blocks, or custom functions may not always be wired correctly.

Example: A variable like birthdate might be declared but not correctly referenced in a custom function or automation step.

Custom Function Errors

Auto-generated functions may contain logical gaps or mismatched input expectations.

Example: A function might expect a variable in one format (e.g., a string) but receive it in another (e.g., an object).

Incomplete Automation Steps

Sometimes, blocks may not be fully configured, leaving placeholders or missing settings.

Example: An API call block might not have the correct headers or endpoint set up.

Unclear Edge Case Handling

The workflow may not account for scenarios like invalid inputs, empty responses, or unexpected user behavior.

Best Practices

Thoroughly Review the Workflow Plan

Before clicking Accept and Build, carefully examine the generated plan, especially variable assignments, function configurations, and block connections.

Test Custom Functions

Use the test interface in the function editor to validate inputs, outputs, and logic. Adjust the code as necessary.

Run the Workflow in Preview Mode

Test the workflow with various inputs to identify gaps or inconsistencies in execution.

Use the Debugger

Analyze runtime logs to ensure all variables and blocks function as intended. Pay close attention to variable updates and outputs.

Iterate on the Build

Treat the generated workflow as a scaffold. Refine and enhance it to meet your specific needs and address any issues.

Final Note

The Workflow Generator is designed to save time and provide a strong starting point, but human oversight is crucial. Always validate the output, refine where necessary, and ensure your workflows are robust and error-free before deploying them in production.

After creating a new AI Agent, you’ll land in the MindStudio Editor. The editor is made of two key areas: the Explorer and the Navigator (commonly referred to as the Workspace Area).

Explorer Tab

On the left you’ll find the Explorer Tab. This is where you'll find all of the resources used to build your AI Agents.

Types of Resources

Data Sources

Upload and vectorize files to leverage Retrieval Augmented Generation (RAG) in your AI workflows.

Learn more about Data Sources →

Functions

Execute JavaScript code in your workflow.

Learn more about Functions →

User Inputs

Interfaces that humans interact with to provide context to the AI Agent.

Learn more about User Inputs →

Workflows

Sequences of automated actions that your AI Agents follow when they are run.

Learn more about Workflows →

Navigator

The large area covering the rest of the Editor if the Main Workspace, also refer to as the Navigator. This area will change depending on what you have selected. By default, the Editor will open on the Main.flow workflow and have the Prompt Tab open.

Additional Global Controls

Top Bar

The Top Bar in the Editor contains the following controls:

• Back Button: Navigates to the Workspace Overview Screen

• Title: Access general AI Agent Settings

• Preview Button: Opens a draft preview of the AI Agent

• Publish Button: Saves and deploys all changes to the AI Agent

Bottom Bar

The bottom bar of the Editor contains the following controls:

• Workspace Name: Click to access workspace settings

• Help & Support: Access video tutorials, documentation, and live support chat

• Collapse Controls: Opens and closes the left and right columns of the Editor

Overview

MindStudio makes it easy to access powerful AI capabilities—without hidden fees or upfront commitments. This guide explains which AI Agents are free to use, how paid agents work, and how to manage your usage balance.

Free AI Agents

MindStudio offers a range of powerful AI Agents that you can use entirely for free. These agents feature workflows designed to efficiently handle more basic tasks. Their simplified processes result in shorter run times, which helps keep the service free for users.

Here are a few examples of free AI agents:

• Research Agents

  Fact-check sources, find similar products, and uncover emails in one click.

• Content Analysis Agents

  Summarize, simplify, or de-bias long articles and videos instantly.

• Content Creation Agents

  Generate tweets, LinkedIn posts, YouTube ideas, and more—on demand.

• Image Generation Agents

  Create icons, illustrations, thumbnails, and other visuals in seconds.

Paid AI Agents

MindStudio also offers AI Agents that leverage complex AI models, advanced workflows, and sometimes even integrate custom data. This additional complexity means these agents are more costly to run.

Here are some examples of tasks run by paid AI agents:

• Running large-scale data analysis

• Using advanced or premium AI models

• Integrating external services and APIs

• Performing long-running or high-volume tasks

To run these agents, you'll need to have a balance in your MindStudio account.

Identifying Free vs. Paid AI Agents

To tell free agents from paid ones, look for a small label beneath each Agent’s Run button. If you see a “Paid Agent” label, running it will deduct from your balance; if there's no label, the Agent is completely free to use.

This indicator appears on both the MindStudio Agents page and the Chrome Extension, so you can confidently browse and try out different Agents knowing exactly which ones are free and which ones require payment.

Example:

In the image below, Deep Research and Research People are marked as Paid Agents, while others like Find Product Alternatives and TL;DR are free to use.

The Prepaid Balance

MindStudio uses a prepaid balance system for running paid AI Agents. Your balance is deducted based on the actual costs of model inference and execution.

Adding Funds to Your Balance

In your Workspace, navigate to Billing > Balance, then select Add Funds to choose an amount and payment method.

Cost of Running Paid AI Agents

The cost per run varies depending on the specific agent and the amount of data processed. Each paid agent displays a cost estimate on its details page, along with an average run time.

No More Usage Fees for Developers

If you're building your own AI Agents in MindStudio, we’ve removed the 2.9% processing fee previously charged on model usage. That means:

• You only pay for model inference costs.

• No API key management required.

• More freedom to experiment and iterate.

Price Breakdown Summary

Need Help?

If you’re unsure whether an AI Agent is free or paid, look for the usage indicator next to the “Run” button or check the Agent's details in the Workspace. For more information, reach out via Support Chat or visit our Quickstart Guide to learn how to build your own AI Agents.

The Start Block is the first part of any workflow in MindStudio. It initializes your AI Agent, defines how and when the workflow begins, and sets up key variables that are used throughout the workflow.

Configurations

Triggers

Define how the workflow is activated. Triggers can be configured to run:

• On-Demand: The default trigger. Requires manual execution.

• Scheduled: At specific times or intervals, such as daily at 8:00 AM.

• Event-Driven: Based on external inputs like API requests.

Launch Variables

Variables that are initialized at the start of the workflow and passed through the entire process. Launch variables can be referenced throughout the workflow using {{variable_name}}.

Scheduling Workflow Runs

The Scheduler is a feature within the Start Block that allows you to set up automated, time-based triggers for your workflow. To enable the Scheduler, change the Triggers > Run Mode configuration from “On-Demand” to “Scheduled”. You may create multiple schedules for a workflow.

Creating Schedules

1. Define the Schedule using natural language to describe how often and when the workflow should run.

   • Examples:

     • "Every day at 8:00 AM"

     • "Every Monday and Wednesday at 3:00 PM"

     • "First day of every month at 9:00 AM"

2. Set the Time Zone based the relative hours you’d like the workflow to run .

3. (Optional) Add Launch Variable arguments using key:value ****pairs.

4. Generate the Schedule by clicking on the Generate Schedule Button

5. Save the schedule.

In this article, we’ll be guiding you through everything you need to know to get started on MindStudio and teaching you to create your first AI Agent.

Prerequisites

Before starting, you will need to create a MindStudio Account.

Getting Started

After creating an account, you'll land on the Workspace Overview Screen, where you can view all published AI Agents. If there are no AI Agents published, then you will see the Getting Started Guide.

Workspace Overview

From the Workspace Overview Screen, you’ll find several controls on the left. From here you can:

• Create new workspaces.

• Access general Workspace Settings.

• Explore Usage Explorer to view AI Agent activity.

• Manage Billing Settings (e.g., budget limits, payment methods).

• Invite team members to your workspace.

• Access developer tools like API Keys, Request Logs, and Documentation.

Building your First AI Agent

In this guide, we’ll build an example AI Agent to find daily tech news, summarize it, and sends an email every morning at 8:00 AM.

Step 1: Create a New AI Agent

To create an AI Agent, click on the Create New Agent button at the top-right. This will create a new Agent and open the MindStudio Editor.

MindStudio Editor Overview

The editor is made of two key areas: the Explorer Tab and the Navigator.

Explorer Tab

This is where you'll find all of the resources used to build your AI Agents. This includes resources like Data Sources, Functions, User Inputs, and Workflows

Navigator (Main Workspace)

The large area covering the rest of the Editor if the Main Workspace, also refer to as the Navigator. This area will change depending on what you have selected. By default, the Editor will open on the Main.flow workflow and have the Prompt Tab open.

Step 2: Write the System Prompt

The System Prompt serves as the AI Agent's core instructions, defining its role, capabilities, and constraints.

You can write a system prompt manually by typing into the blank space below, or you can click on the Generate Prompt button at the bottom left to have the Prompt Generator to write the prompt for you.

Using the Prompt Generator

• Click on the Generate Prompt button at the bottom left of the prompt area. This will open the Prompt Generator Modal.

• Using natural language, enter a brief description of what your AI Agent is supposed to do. For this build, we can type something like:

  Copy

  Find daily technology news, summarize it, and send an email at 8:00 AM.

• After entering your description, click the Generate button. The system will automatically create a structured prompt based on your input.

• Carefully review the generated prompt to make sure it aligns with your requirements. You can edit the prompt directly in the editor if adjustments are needed.

Once your System Prompt is written, navigate to the Automations tab.

Step 3: Design a Workflow in the Automations Tab

The Automations Tab is where you design the workflow for your AI Agent. This section allows you to define the sequence of actions your AI Agent will follow.

Example Workflow: Daily Tech News AI Agent

1. Start Block:

   • Schedule the workflow to run at 8:00 AM daily.

   • Define a launch variable called topic.

2. Google Search Block:

   • Use a function to perform a Google search.

   • Reference the topic variable as the search query.

   • Save the result as a new variable, google_result.

3. Text Generation Block:

   • Prompt AI to summarize the google_result.

   • Use the prompt:

   Copy

   Summarize the following information: {{google_result}}

4. Terminator End Block w/ Email Notification:

   • Enable Email notifications to send the summary to the specified email.

Step 4: Choose an AI Model

MindStudio provides access to over 50 AI models from leading providers, allowing you to tailor your workflows for a variety of tasks and use cases.

1. Access Model Settings

   • Navigate to the Model Settings Tab in your workflow editor.

2. Choose a Model

   • Select from over 50 AI models available.

   • NOTE: Some models may be locked and can be unlocked by adding a payment method in your Billing Settings.

Configuring Model Settings

There are two configuration options for AI models:

1. Temperature

   • Controls the randomness of responses:

     • Higher Temperature: More creative, but less predictable.

     • Lower Temperature: More consistent, but less diverse.

2. Max Response Size

   • Defines the maximum size of the model’s response in tokens.

   • Example: 400 tokens ≈ 300 words.

Using Multiple AI Models Within a Workflow

Every block that uses AI has the ability to override the underlying models settings and use its own unique AI model and configuration. This allows you to use different models in the workflow based on the task required at each specific step.

Step 5: Test and Evaluate Your Workflow

Testing and evaluation are critical steps to ensure your AI Agent performs as expected. MindStudio provides tools to debug, validate, and optimize your AI Agents.

How to Check for Errors in Your Workflow

• Navigate to the Errors Tab to view any issues, such as misspelled variables or misconfigured blocks.

• Click on an error in the Errors Tab to highlight the problematic block in the workflow.

• Adjust configurations or correct spelling to resolve the issue.

• Once fixed, errors will disappear from the tab.

Use Evaluations Tab to Generate Test Cases

• Open the Evaluations Tab to create test scenarios for your workflow.

• Click the Autogenerate button to generate test cases automatically. You can provide additional context for test case generation, such as specifying genres, topics, or scenarios.

  Example: Generate test cases for topics like technology, politics, or health.

• Specify how many test cases you want to generate.

• Review the generated cases to ensure they cover a variety of inputs

• Use the Run button in the top-left to execute the workflow multiple times with the test cases.

• After execution, review the results in the right-hand column. Check whether the outputs align with expected behavior.

Step 6: Publish Your AI Agent

Once your AI Agent has been tested and evaluated to ensure it functions correctly, the next step is to publish it. Publishing makes your AI Agent available for use and integration.

Steps to Publish Your AI Agent

1. Set Metadata

   • Navigate to the Explorer Tab and click on the root menu.

   • Add the following details:

     • Name: Give your AI Agent a descriptive and unique name.

     • Description: Provide a short explanation of what your AI Agent does.

   • Go through other sections and configure other settings.

2. Publish the Workflow

   • Click the Publish button to finalize your AI Agent.

   • Once published, the AI Agent will be accessible via the Published Tab in your workspace. Click on the confirmation message to view your AI Agent.

3. Integration Options

   • After publishing, you can integrate your AI Agent with external tools and platforms, such as:

     • Zapier: Automate workflows across apps.

     • Make: Build advanced automation scenarios.

     • API: Use the provided documentation to connect via custom APIs.

4. Test Request Integration

   • Use the Create Test Request feature to validate API calls.

   • Configure the API request by:

     • Selecting a workflow.

     • Providing launch variables (e.g., topic = "daily tech news").

   • Run the test and review the logs for successful execution.

5. View the AI Agent in the Workspace Overview

   • View the newly published AI Agent under the Home Tab.

   • Confirm it is listed alongside other published Agents.

Conclusion

Congratulations on creating and publishing your first AI Agent in MindStudio! We’ve covered everything from writing system prompts and building workflows to testing, debugging, and publishing. MindStudio is a powerful and versatile platform, enabling you to create AI Agents tailored to specific needs.

Remember, this guide is just the beginning. The possibilities with MindStudio are endless, and we’re excited to see what you’ll build. If you ever need help, our Support Chat is just a click away.

Prompts are the foundation of how your AI Agent understands and executes tasks. A well-written prompt ensures your AI Agent delivers precise and meaningful results.

What is a Prompt?

A prompt is a set of instructions that tells the AI Agent what to do. It provides the context and parameters for generating the desired output. Just like giving someone directions, a good prompt ensures the AI knows exactly what to deliver. In MindStudio, prompts can be used at two levels: System Prompts and Block Prompts.

System Prompt

The System Prompt appears in the Prompt Tab of a workflow file. It serves as the AI Agent's core instructions, defining its role, capabilities, and constraints and acts as the foundation, guiding how the AI behaves throughout the workflow.

When you write a system prompt, you’re establishing the AI's "role" and general approach to tasks. Every action in the workflow will follow this overarching guidance unless overridden by specific block prompts.

You can write a system prompt manually by typing into the blank space of the Prompt Tab, or you can click on the Generate Prompt button at the bottom left to have the Prompt Generator to write the prompt for you.

Using the Prompt Generator

1. Click on the Generate Prompt button at the bottom left of the prompt area. This will open the Prompt Generator Modal.

2. Using natural language, enter a brief description of what your AI Agent is supposed to do.

   1. Example:
3. After entering your description, click the Generate button. The system will automatically create a structured prompt based on your input.

4. Carefully review the generated prompt to make sure it aligns with your requirements. You can edit the prompt directly in the editor if adjustments are needed.

Block Prompt

A block prompt is used for specific tasks within the workflow. While the system prompt provides overall guidance, a block prompt gives detailed instructions for a particular step.

Example of a Block Prompt (within a Generate Text block):

Here, the block prompt tells the AI to focus on summarizing a specific piece of information (e.g., the result of a Google search). This allows you to customize how the AI performs at different stages of the workflow.

How System Prompts and Block Prompts Work Together

Key Aspects of Prompt Writing

A good prompt is clear, specific, and provides enough context for the AI to understand what’s needed.

Key aspects of an effective Prompts:

• Clarity: Use straightforward language to avoid confusion. Clearly define the AI Agent's purpose and main tasks

• Specificity: Be detailed about what you want. Specify any limitations or guidelines that your AI Agent should follow when executing its task

• Context: Explain the purpose or provide a scenario to guide the AI. Context can be provided via direct instruction or by calling Variables within a prompt.

• Output Format: Outline the expected format or structure of its output.

Examples of Good Prompts vs. Bad Prompts

Using Markdown to Write Prompts

Markdown is a lightweight formatting language that can be used to structure prompts, making them clearer and easier for the AI to interpret. By using Markdown, you can organize your instructions in a readable and visually structured way.

Why Use Markdown in Prompts?

1. Clarity: Break down complex instructions into digestible sections.

2. Readability: Using formatting like headers, bullet points, and code blocks to make prompts easier to follow.

3. Consistency: Ensure your prompt has a standardized structure, especially for complex tasks.

4. Debugging: Allows for easier debugging and refinement of prompts.

Common Markdown Elements for Prompts

Here are some common Markdown features and how to use them:

1. Headers

   • Purpose: Use headers to introduce sections or tasks.

   • Variations:

     • # for H1

     • ## for H2

     • ### for H3

     • #### for H4

   • Example:
2. Bullet Points

   • Purpose: Break down requirements into bullet points for clarity.

   • Example:
3. Numbered Lists

   • Purpose: Use for step-by-step instructions.

   • Example:
4. Code Blocks

   • Purpose: Use code blocks for examples or templates.

   • Example:
5. Bold and Italics

   • Purpose: Emphasize key instructions.

   • Example:

Example System Prompt Using Markdown

Auto-Enhance Prompts

The Auto-Enhance feature helps polish and refine your prompts after you've made modifications. It analyzes your prompt for clarity, formatting, and potential ambiguities, then suggests improvements while maintaining your original intent.

How Auto-Enhance Works

1. Access the feature: Click the "Enhance" button in the bottom bar or use the keyboard shortcut Option + K

2. Review suggestions: The system presents an enhanced version of your prompt with improvements in:

   • Grammar and spelling

   • Markdown formatting consistency

   • Clarity and precision of instructions

   • Structure and organization

3. Accept or reject changes: Choose whether to implement the suggested improvements

This feature works both in the main Prompt Tab and in fullscreen Send Message blocks, making it easy to maintain high-quality prompts throughout your workflow.

MindStudio provides access to over 50 AI models from leading providers, allowing you to tailor your workflows for a variety of tasks and use cases.

Select an Underlying AI Model

The underlying model is the AI model that is inherited by the rest of the workflow. It’s configuration will be used as the default model for all blocks that use AI.

1. Access Model Settings

   • Navigate to the Model Settings Tab in your workflow editor.

2. Choose an AI Model

   • Click on the Model Card to view information about the AI Model.

   • NOTE: Some models may be locked and can be unlocked by adding a payment method in your Billing Settings.

3. Click the Use Button to confirm your selection

4. (Optional) Click the heart icon to add preferred AI Models to your Favorites section

Configuring Model Settings

There are two configuration options for AI models:

1. Temperature

   • Controls the randomness of responses:

     • Higher Temperature: More creative, but less predictable.

     • Lower Temperature: More consistent, but less diverse.

2. Max Response Size

   • Defines the maximum size of the model’s response in tokens.

   • Example: 400 tokens ≈ 300 words.

Best Practices for AI Model Configuration

• Start the slider in the center for temperature and adjust based on your needs.

• Use larger response sizes for detailed outputs and smaller sizes for concise ones.

Using Multiple AI Models Within a Workflow

Every block that uses AI has the ability to override the underlying models settings and use its own unique AI model and configuration. This allows you to use different models in the workflow based on the task required at each specific step of the workflow.

Considerations When Selecting an AI Model

Choosing the right AI model is critical to ensuring your workflow meets performance, cost, and quality requirements. MindStudio provides a variety of models with different capabilities, trade-offs, and configurations. When selecting a model, it's often necessary to balance these considerations to align with your workflow's goals and constraints.

Price

AI models come with varying pricing structures based on usage, typically measured in tokens for prompt and response.

Use cost-effective models for high-volume, repetitive tasks (e.g., bulk summarization). Opt for premium models only when high output quality is critical.

Latency

Latency refers to the time the model takes to generate a response. Low-latency models are essential for real-time or interactive workflows.

Prioritize low-latency models for use cases like chatbots or live applications. For non-time-sensitive workflows (e.g., scheduled reports), higher latency models with better quality may be acceptable.

Output Quality

Different models vary in their ability to generate coherent, creative, or factual responses. Output quality depends on the model’s training and capabilities.

Choose advanced models for nuanced tasks like legal summaries or creative writing. Use simpler models for straightforward tasks like data extraction.

Context Window

The context window determines the maximum amount of text the model can process at once. Larger context windows are essential for tasks involving lengthy inputs.

Use models with large context windows for summarizing lengthy documents or analyzing extensive datasets. For shorter inputs, a smaller context window may suffice and reduce costs.

Learn About Specific AI Models

The Automations Tab is where you design the workflow for your AI Agent. This section allows you to define the sequence of actions your AI Agent will follow.

Workflow Blocks Reference

What is a Workflow?

Workflows are sequences of automated actions that your AI Agents follow when they are run.

Designing Workflow Automations

After opening the Automations Tab, you’ll see the Automations Canvas which displays all of the individual actions that your workflow will execute when run. These are represented by Blocks on the canvas.

Best Practices for Designing Workflows

• Plan Before You Build: Outline the steps your workflow will perform.

• Use Helpful Variable Names: Name variables descriptively to make the workflow easy to debug.

• Test as You Go: Regularly test each block to identify and resolve issues early. To test out the specific output of blocks that use AI, you may use the Profiler to compare results.

Automations Canvas Controls

The canvas is infinitely scrollable in all directions. At the bottom of the canvas, you’ll find several controls to help you navigate and annotate the canvas as you design your Workflow.

Panning Around the Canvas

Using a mouse:

• Scroll to pan up and down the canvas.

• Shift + Scroll to pan left and right across the canvas.

Using the Pan Tool:

• Click the Pan Tool icon or use the H hotkey to activate the Pan Tool.
• Click and drag in any direction to pan around the canvas. NOTE: You will not be able to select blocks while the Pan tool is activated.

• Click the Select Tool icon or use the V hotkey to deactivate the Pan Tool.

Zooming In & Out

Using a mouse:

• On Mac, use CMD + Scroll to zoom in and out.

• On PC, use CTRL + Scroll to zoom in and out.

Using the Zoom Controls:

• Click Zoom In icon or use the + hotkey to zoom in.
• Click Zoom out icon or use the - hotkey to zoom out

Centering the Canvas

If you navigate away from your blocks and can no longer find them, you can reset and center all of the blocks on the canvas by clicking the Reset view icon or using the R hotkey

Auto Arrange Blocks

If your workflow becomes long and complex, you may want to consider tidying it up using Auto arrange. This tool will align all blocks vertically on the canvas.

Adding Notes to the Canvas

You can annotate the canvas using the Note Tool to add text notes or label groups of blocks on.

Using the Note Tool Controls:

• Click on the Note tool icon to create a new note
• Select the color of your note

• Click on the note and add text content to your note

• Add an optional label to the top of your note. If no text is added in the label, then it will not show a label.

• Click and drag the anchors found at each corner of the note to adjust the note size.

Add a New Block to the Workflow

Between Existing Blocks:

• Select the + button between to connected blocks to open up the Block Menu.

• Select the block that you’d like to add.

• The new block will be automatically connected between the two blocks.

Anywhere on the Automations Canvas:

• Right-click anywhere on the canvas to open the Block Menu.

• CTRL + Click anywhere on the canvas to open the Block Menu.

Deleting a Block

• Click on the delete icon at the top-right corner of any block.

• Confirm the delete action.

Key Components of a Workflow

The Start Block

This block initializes the workflow. It can be triggered on demand or on a defined schedule, and allows you to define launch variables, which provide dynamic values that are passed through the workflow.

Workflow Blocks

There are many kinds of blocks that you can add to your workflow. All blocks will have different configuration options depending on the block that you select.

Types of blocks include:

• AI Tools: Generate text, generate images, analyze image, etc.

• Context Blocks: Gather context for the AI. Context is saved to a variable.

• Routing Blocks: Add conditional branches to route the workflow in various ways. </aside>

Terminator Blocks

This block marks the end of the workflow. It has customizable end behavior, such as sending email notifications or returning a structured output.

Defining and Calling Variables

Variables in MindStudio are dynamic placeholders that store data during workflow execution. They allow you to pass information between blocks and workflows seamlessly.

Defining Variables

1. Launch Variables: These are defined in the Start Block of your workflow.

2. Runtime Variables: Some blocks, such as Generate Text Blocks or User Input Blocks, can generate new variables while the workflow is running. For Example, after performing a Google Search, the block can store the results in a variable called google_result.

Calling Variables

To use a variable in any block, reference it using double curly braces: {{variable_name}}.

Generate Text prompt example calling a variable:

Dynamic Variables in MindStudio enable workflows to generate and modify user inputs in real time, allowing for more adaptive and interactive experiences. This feature is particularly useful for scenarios where follow-up questions depend on previous responses, such as product recommendations, lead generation, and troubleshooting agents.

By leveraging variables, arrays, and Handlebars helpers, Dynamic Inputs allow workflows to progressively collect information and adjust questions based on user input.

How Dynamic Inputs Work

Dynamic Inputs use variables to store and retrieve responses, allowing the workflow to adjust its next question dynamically. This pattern is useful in AI Agents that require multiple layers of questioning before generating a final result.

Key Elements

1. Inputs Array – myVariable[] - Stores inputs generated into an array.

3. (optional) Handlebars helpers - Used to quickly find items in the array variable.

Building a Workflow with Dynamic Inputs

1. Collect Initial User Information

Start by gathering basic user input using a User Input Block. This might include general details like name, contact information, or the primary topic of inquiry.

Example: "What part are you looking for?"

The response is stored in a variable, such as partRequest.

2. Generate a Follow-Up Question

Once the user provides an initial answer, the workflow generates a follow-up question based on the input. This follow-up question is:

• Added to the Questions Array variable: questions[].

• Displayed dynamically as the next prompt using: {{lastItem questions}}.

Example:

• The user inputs "fire sprinkler".

• The workflow generates "What type of sprinkler are you looking for?" and stores it in questions.

3. Store User Responses

Each response is stored in the Answers Array Variable (answers[]), ensuring all collected data is saved for final processing.

Example:

• The user selects "Pendant", which is added to answers.

4. (optional) Logic Block for Additional Questions

The Logic Block determines:

• If more information is needed, the workflow loops back, dynamically generating another follow-up question.

• If sufficient responses are collected, the workflow moves forward to generate the final result.

Example:

• Second Follow-Up Question: "What temperature rating do you need?"

• Third Follow-Up Question: "What finish or color scheme?"

5. Generate the Final Result

Once all necessary inputs are collected, the workflow processes the information and generates a response based on the accumulated answers.

Example Output: "You are looking for a brass pendant fire sprinkler with a 135°F temperature rating."

Example: Fire Sprinkler Part Finder

This workflow dynamically adjusts to refine user requests:

Each follow-up question is stored in questions[], and the corresponding user input is stored in answers[].

Use Cases for Dynamic Variables

Dynamic Inputs open new possibilities for AI-driven workflows:

• Lead Qualification – Tailor follow-up questions based on user responses to better qualify leads.

• Product Recommendations – Ask clarifying questions to suggest the best product options.

• Troubleshooting Assistants – Adjust questions dynamically based on user-reported issues.

Conclusion

• Dynamic Variables leverage arrays to adjust varaibles in real time.

• The Logic Block determines when to stop questioning and proceed with generating results.

• Handlebars helpers like {{lastItem}} retrieve the most recent question for dynamic presentation.

This pattern is a powerful way to build interactive and intelligent AI Agents that adapt to user needs dynamically. While it requires a deeper understanding of arrays and variable management, the flexibility it provides makes it a game-changer for many AI-driven workflows.

The Generate Text Block sends a text prompt to an AI model and returns the AI Model’s response.

Configurations

Prompt

Define the instructions sent to the AI model. Use {{variables}} to make the prompt dynamic and context-aware.

Response Behavior

Choose how the AI response is handled:

• Display to User (Default): Shows the response to the end user.

• Assign to Variable: Creates a new variable and saves the AI model’s response to it. Enter a variable_name ****to store the response for later use in the workflow.

Output Schema

Choose the format of the AI response:

Text (Default):

Outputs plain or markdown-formatted text, suitable for display or emails.

JSON:

Outputs the response in JSON format, ideal for structured data outputs required for integrations or further processing via code.

Example Schema:

CSV:

Outputs data in CSV format, ideal for tabular data or spreadsheets.

Example Schema:

For both JSON and CSV outputs, you can explicitly define the output structure by providing a sample output schema.

Model Settings

The default setting is inherited from the underlying model configured in the Model Settings Tab. You can override this setting by choosing and configuring a different AI Model specifically for this block.

• Model Selection: Choose a different model if needed for the block.

• Temperature: Adjust randomness in the response:

  • Lower values = more predictable, consistent results.

  • Higher values = more creative, varied responses.

• Response Length: Limit the output size in tokens to suit your needs.

Using Conditional Logic in Your Prompt

The Generate Text Block supports conditional logic to dynamically adjust the text prompt based on workflow variables. This is done using {{#if}}, {{else}}, and {{/if}} statements.

How to Use Conditionals:

Basic Syntax:

Example:

Nested Conditionals:

Combine multiple {{#if}} statements for complex logic:

A Data Source in MindStudio is a repository of files that can be queried by AI workflows to supplement the AI’s responses with domain-specific knowledge. By integrating custom data into your workflows, Data Sources enable Retrieval Augmented Generation (RAG)—a method where the AI retrieves relevant information from external data sources to generate more accurate and contextually aware outputs.

When you upload files to a Data Source, they are converted into a vector database. This database allows the AI to query your custom data, retrieve relevant context, and incorporate it into its responses.

What is Retrieval Augmented Generation?

RAG combines AI models with an external knowledge base to enhance the AI's capabilities. Instead of relying solely on pre-trained data, the AI retrieves relevant information from the Data Source and integrates it into its response.

How RAG Works

Knowledge Retrieval

Before generating a response, the system retrieves relevant information from external sources, such as a knowledge base, database, or documents. This is typically done using vector embeddings and similarity search to find the most relevant pieces of information.

Augmented Generation

The retrieved knowledge is used as input or context for the AI model during the generation phase. This ensures that the output is informed by domain-specific information, increasing accuracy and relevance.

Why Use RAG?

RAG combines the strengths of retrieval-based systems and generative AI. Instead of relying solely on the model's pre-trained knowledge, it integrates current and specific knowledge at runtime. RAG also reduces hallucinations (making up information) since the generated text is guided by real, retrieved data. Lastly it can handle large, external knowledge bases without the need to retrain all of the information directly as context for the model.

Using Data Sources

To make the AI use a Data Source, you must include a Query Data Source Block in your workflow. This block allows you to instruct the AI on how to query the uploaded data and integrate relevant information into its responses.

Example Workflow:

1. Upload Product Manuals: Create a Data Source with PDFs of your product documentation.

2. Query Data Block: Use the block to retrieve specific sections based on user queries, such as “How do I reset my device?”

3. Generate Text Block: Combine the retrieved data with the AI’s language capabilities to deliver a clear and accurate response.

Managing Data Sources

Create a Data Source

1. Navigate to the Data Sources Folder from the Explorer Tab

2. Click the (+) icon to create a new Data Source.

3. Configure the Data Source with a name and description

4. Upload your files.

Supported File Types

• PDF (.pdf)

• CSV (.csv)

• Word Document (.docx)

• Excel Spreadsheet (.xlsx)

• Text File (.txt)

• HTML File (.html)

Limits for Uploads:

• 500 MB per file

• 5 Million words per file

• Maximum of 150 files per Data Source

Managing Data Sources

Viewing Individual Files

Once a Data Source is created, you can view and explore its individual files to verify the data or ensure that the uploaded content has been processed correctly. Double-clicking on a Data Source opens a detailed view of its contents. When you open a file within a Data Source, the interface provides multiple tabs for exploring different aspects of the data:

• Preview: Displays the file preview, giving you a quick way to review the original content. This is useful for verifying that the file has been uploaded correctly.

• Extracted Text: Shows the plain text extracted from the file, stripped of any formatting. This is the data the AI will query during workflows.

• Raw Chunks: Displays the segmented "chunks" of text that the system uses to create the vector database. Chunks ensure the data is broken down into manageable, queryable parts.

• Raw Vectors: Shows the numerical vector representations of the data. These vectors are used by the AI for retrieval during queries.

Force Reload a Data Source

Force reloading ensures that the Data Source reflects the latest changes after you’ve uploaded new or updated files.

1. Open the Data Sources Folder in the Explorer Tab.

2. Right-click the Data Source you want to reload.

3. Select Force Reload from the drop-down menu.

4. The Data Source will remove all previous data not currently included and update accordingly.

Rename a Data Source

You can rename a Data Source through the Data Source Editor or directly from the Explorer Tab.

From the Data Source Editor:

• Open the Data Sources Folder.

• Select the Data Source to rename.

• In the editing screen, update the name in the Name field.

• Changes are saved automatically.

From the Explorer Tab:

• Open the Data Sources Folder.

• Right-click the Data Source.

• Select Rename and enter the new name in the form field.

Delete a Data Source

1. Open the Data Sources Folder in the Explorer Tab.

2. Right-click the Data Source you want to delete.

3. Select Delete from the drop-down menu.

4. Confirm the deletion.

Anatomy of a Data Source

The Data Source Details panel provides an overview of a Data Source's configuration and its current state. This section also includes the Query Tester, a tool to verify that your Data Source is properly loaded and functioning as expected.

Data Source Details

This section displays essential information about your Data Source, helping you understand its status and content.

• Name: The name of the Data Source. This can be customized to make it easier to identify.

• Description: A brief optional description of the Data Source. If no description is provided, it will display as "No description."

• Status: Indicates whether the Data Source is loaded and ready for use. Includes a word count progress bar (e.g., "0.1k/5m words") to show how much data has been loaded relative to the word limit.

• Documents: The number of files or documents currently included in the Data Source.

• Words: The total number of words across all documents in the Data Source.

• Vectors: The total number of vectorized entries in the Data Source. Vectors are created during the conversion of documents into a searchable format.

Query Tester

The Query Tester is a built-in tool to test queries and validate that your Data Source is working correctly.

• Query Input: A text box where you can type your query. This is how you simulate an AI prompt or user question to test the Data Source’s retrieval capabilities.

• Run Query Button: Clicking the play button sends the query to the Data Source and returns results, allowing you to verify the response.

Example Use Case:

If your Data Source contains product manuals, you can test a query like:

How do I reset my device?

Data Sources Best Practice

Data Sources enhance your AI workflows by integrating custom knowledge. Follow these best practices to maximize their effectiveness:

Organize Your Data

Group related files in the same Data Source for efficient queries and use clear, descriptive names like "Product_Manuals_2024."

Prepare Your Files

Clean files to ensure relevance and accuracy. Use supported formats like PDF, CSV, DOCX, and stay within upload limits (500 MB per file, 150 files per Data Source).

Focus Queries

Write specific prompts to guide the AI in retrieving the right information. Combine queries with variables for dynamic, context-aware responses.

Keep Data Sources Updated

Regularly refresh content when information changes and use Force Reload to apply updates after modifying or adding files.

Optimize Performance

Segment large datasets into smaller, focused Data Sources and avoid overloading Data Sources to maintain query accuracy and speed.

The Analyze Image Block processes and analyzes an image URL based on provided text instructions. It generates a text response based on the analysis that is assigned to a variable.

Configurations

Prompt

Provide text instructions for analyzing the image. Use {{variables}} to make the prompt dynamic based on a previous step in the workflow.

Example Prompt:

Identify the objects in the image and describe their arrangement. Focus on colors and sizes.

Image URL

Specify the URL of the image to be analyzed. Use a {{variable}} to dynamically reference image URL generated or fetched earlier in the workflow.

Output Variable

Creates a new variable and saves the generated test response to it. Enter a variable_name to store the response for later use in the workflow.

Model Settings

Choose the AI Model that you’d like to use for the image analysis. You may adjust the model’s Temperature and Max Response Size.

Writing Image Analysis Prompts

Creating a well-crafted prompt is essential for accurate and meaningful image analysis. Follow these guidelines to ensure your prompts guide the AI model effectively:

• Be Clear and Specific: Clearly describe what you want the AI to analyze or identify in the image. Focus on specific elements or features relevant to your use case.

  Example: "Identify the main objects in the image, including their colors, sizes, and positions.”

• Define the Objective: Explain the purpose of the analysis to provide context for the AI. This helps tailor the response to your needs.

  Example: "Analyze the image and describe how the objects are arranged for interior design recommendations.”

• Break Down Complex Instructions: Use simple, step-by-step instructions when requesting detailed analyses.

  Example: " 1. List all objects visible in the image. 2. Describe their positions relative to each other. 3. Highlight any unusual or standout features.”

• Focus on Key Details: Avoid asking for unnecessary information to keep the analysis concise.

  Example: "Describe the objects on the table, focusing on their materials and colors.”

• Include Contextual Cues: Provide additional context for better results, such as the setting, purpose, or focus of the image.

  Example: "This image shows a park scene. Identify all visible plants, trees, and animals.”

The Display Content Block allows you to present messages or outputs directly to users in your workflows. It is used for delivering dynamic content with support for markdown formatting.

Configurations

Message

Define the message you want to display to the user. The Display Content Block supports markdown formatting, allowing you to structure text, add emphasis, or include links for a polished presentation. Use {{variables}} to make the message dynamic.

Learn about variables →

Learn more about Markdown formatting →

Example Message:

## Welcome to our app!
Here are your **next steps:**
1. Complete your profile.
2. Explore the features of {{productName}}.
3. Contact us [here](<https://example.com/support>) for assistance.

Learn more about markdown → [PAGE: Writing Prompts with Markdown].

Display an Image

![Alt Text]({{myImageVariable}})

Display Audio Player

<audio src="{{myAudioVariable}}" controls /></audio>

The User Input Block lets your workflows to collect data directly from end users. Unlike AI-focused blocks, this block presents the forms or interfaces where users can provide input. When New User Inputs are automatically added to the User Inputs folder in the left-side Explorer tab.

What is a User Input?

A User Input in MindStudio is a form-based interface that collects data directly from your end users. This data serves as context or input for workflows, and is stored as a variable and serves as context for downstream processes and AI-powered actions within your workflow .

When New User Inputs are created, they are automatically added to the User Inputs folder in the left-side Explorer tab.

Add Existing User Inputs

1. Click on the + button at the bottom of the User Inputs Configuration Panel.

2. In the modal, choose from created User Inputs to add to the block.

3. (Optional) Click and drag User Inputs up and down to reorder them.

4. Click the Add button to confirm your choices.

Create A New User Input

From the User Input Block:

1. Click on the (+) button at the bottom of the User Inputs Configuration Panel.

2. In the modal, click the Create New… button at the bottom left. A new User Input will be created in Explorer within the User Inputs folder, and will be automatically added to the block.

3. Configure the user input.

From the Explorer:

1. Hover over the User Inputs folder, then click on the + button to the right of the folder. A new User Input will be created in Explorer within the User Inputs folder.

2. Configure your User Input.

3. After configuring the User Input, you will need to add the User Input to the User Input Block.

Edit A User Input

1. Open the User Inputs folder from the Explorer tab.

2. Click on the User Input that you’d like to modify

3. Changes to configurations are automatically saved.

Duplicate a User Input

1. Open the User Inputs folder in the Explorer tab.

2. Right-click on the User Input to duplicate.

3. Select the Duplicate option.

The duplicated User Input displays in the User Input folder using the root name of the original with a number following it.

Delete a User Input

Deleting a User Input removes it from all User Input Blocks across all Workflows.

1. Open the User Input folder in the Explorer tab.

2. Right-click on the User Input to delete.

3. Select the Delete option from the drop-down menu.

4. Confirm your deletion.

Anatomy of a User Input

Type

The type of User Input determines its functionality and format. See the table below for a full list of User Input types.

Variable

The variable_name is a unique identifier for the User Input. It is used to reference the collected data in downstream workflow blocks. Use a variable name that is unique to the data being collected. (Example: customer_goal , client_industry , first_name )

Configurations

Configurations define how the User Input behaves and what options are available. These settings vary depending on the input type.

Featured Image

An optional image displayed above the input field when presented to the end user. It enhances visual appeal and provides context to your end user.

Test Value

Pre-fills the input field with a sample response when previewing workflows. This simplifies front-end testing when viewing a draft preview of an AI Agent.

MindStudio is a powerful and versatile platform that empowers you to create all kinds of AI Agents tailored to your unique needs. Whether you're streamlining operations, enhancing customer experiences, or boosting productivity, MindStudio provides the tools to design these intelligent solutions that fit neatly into your existing business or tech stack.

This article explores the diverse range of AI Agents you can build with MindStudio, showcasing real-world examples to spark inspiration and help you unlock the full potential of AI in your workflows.

Conclusion

The examples we've explored are just a small glimpse into what’s possible with MindStudio. From automating routine tasks to crafting sophisticated AI-powered solutions, the possibilities are endless. MindStudio puts the power of AI at your fingertips, allowing you to design workflows that are as unique as your bespoke business needs.

These examples are a starting point—what you build is entirely up to you. The possibilities are vast, and the direction you take is yours to define. With MindStudio, you have the flexibility and capability to build exactly what your business needs to thrive in the age of AI.

Variables in MindStudio are dynamic placeholders that store data during workflow execution. They allow you to pass information between blocks and workflows seamlessly.

Example:

• Variable Name: userName

• Usage: "Hello, {{userName}}! Welcome to our app."

Creating Variables

Variables are created automatically in MindStudio whenever:

• A User Input collects data.

• A block generates an output (e.g., Generate Text Block, Analyze Image Block).

• You manually define them in the Start Block.

Types of Variables

Launch Variables

These are defined in the Start Block of your workflow. Values for these variables are passed in as arguments when a workflow is run via API or via the Run Workflow block.

Runtime Variables

Some blocks, such as Generate Text Blocks or User Input Blocks, assign values for the variable while the workflow is running. For Example, after performing a Google Search, the block can store the results in a variable called google_result.

Calling Variables

To use a variable in any block or prompt, reference it by enclosing the variable name in double curly braces: {{variable_name}}.

Example Calling Variables in a Generate Text Block:

Summarize the article titled "{{articleTitle}}". 

In the summary, make sure to write about the following topic: 
{{topic}}

Extracting a Value from a JSON Structure

MindStudio provides tools for extracting specific values from JSON objects using the JSON Path syntax and the get helper. This allows workflows to handle and manipulate structured data with precision, making them more dynamic and adaptable.

get Helper - Query JSON Variables

The get helper allows you to query JSON variables for specific values using JSON Path expressions. This feature is especially useful when working with nested or complex JSON structures.

Syntax:

{{get myVariable "$.path.to.value"}}

Example 1: Extract a Nested Value:

Given the following JSON assigned to myJsonVariable:

{
    "user": {
        "name": "Alice",
        "details": {
            "email": "alice@example.com"
        }
    }
}

Use this to extract the email address:

{{get myJsonVariable "$.user.details.email"}}

Output: alice@example.com

Example 2: Extract the First Item in an Array:

Given the following JSON:

{
    "items": [
        {"id": 1, "name": "Foo"},
        {"id": 2, "name": "Bar"}
    ]
}

Use this to extract the name of the first item:

{{get myJsonVariable "$.items[0].name"}}

Output: Foo

Example 3: Extract Multiple Values:

JSON Path also allows for querying multiple elements. Given the following JSON:

{
    "items": [
        {"id": 1, "name": "Foo"},
        {"id": 2, "name": "Bar"}
    ]
}

{{get myJsonVariable "$.items[*].name"}}

Output: ["Foo", "Bar"]

Best Practices When Extracting JSON

1. Use a JSON Path Tester: Tools like JSONPath Online Evaluator can help you refine and test your JSON Path queries.

2. Validate JSON Structure: Ensure your variable contains valid JSON data before attempting to extract values.

3. Handle Missing Values: Include fallback logic in your workflow to handle cases where the expected path does not exist in the JSON.

Using Handlebars Templating

MindStudio leverages the Handlebars templating language to make working with variables intuitive and powerful. Handlebars allows you to include, manipulate, and conditionally render data directly in your prompts, outputs, and logic.

Conditional Logic

Handlebars supports if-else logic for dynamic outputs:

{{#if userName}}
Hello, {{userName}}! How can we assist you today?
{{else}}
Hello! Please log in to get started.
{{/if}}

For a full list of expressions, see Handlebars Documentation.

Special Handlebars Methods in MindStudio

In addition to standard Handlebars features, MindStudio introduces two special methods for advanced functionality:

{{json varName}}

Converts a JSON object into a string format.

Example:

If userProfile contains:

{
    "name": "John",
    "age": 30
}

Usage:

{{json userProfile}}

Output:

{"name":"John","age":30}

{{sample varName number token}}

Extracts a portion of the variable's content based on specified parameters.

• Parameters:

  • varName: The variable to sample.

  • number: The number of items (e.g., lines, words, or letters). If negative, starts from the end.

  • token: The type of unit to extract (line, word, or letter).

• Examples:

  • Extract the first 5 words:

    Copy

    {{sample textVar 5 word}}

  • Extract the last 3 lines:

    Copy

    {{sample textVar -3 line}}

  • Extract the first 10 letters:

    Copy

    {{sample textVar 10 letter}}

The Text to Speech Block converts written text into voice-over audio, allowing you to dynamically generate audio content based on workflow inputs.

Configurations

Text

Define the text you want to convert to audio. Use {{variables}} to dynamically insert text you’d like converted into audio.

Output Variable

Creates a new variable and saves the audio file URL to it. Enter a variable_name to store the response for later use in the workflow.

Model Settings

Choose the AI Model you’d like to use to generate the audio. Different models will have unique settings to adjust the output’s characteristics. Review the selected model’s specific options for a full list of configurations.

The User Context block utilizes AI to generate a series of questions to the user based on certain parameters.

Configuration

Prompt

Define the additional data you want to collect from the user. Like the Generate Text block, the prompt serves as a message to the AI, directing it to produce a specific output. In this instance, that output is a set of interview-style questions designed to gather more context from the user.

Interview Depth

Specify the level of context gathered by the AI. This setting will inform how probing the questions are and how comprehensive the final report will be.

Three options are provided:

1. Quick (default)

2. Medium

3. Thorough

Max Questions

Set the amount of questions to be asked. By default, the max amount of questions will be set to 5.

Variable Name

Save the responses to a variable that can be used in other parts of the workflow. Example: My_Var

Output Format

Select the output format for the response data:

1. Text (Default)

2. JSON

The Extract Text from File Block allows you to extract text content from a file provided via a URL. This block is perfect for workflows that require processing text from uploaded files or external sources.

Configurations