Getting Started with Miniloop: A Complete Guide

Welcome to Miniloop! This comprehensive guide will walk you through creating your first AI-powered workflow from scratch. By the end of this tutorial, you'll understand how to build, test, and deploy automated data processing workflows using Miniloop's AI-assisted platform.

What is Miniloop?

Miniloop is an AI-powered workflow automation platform designed for processing data at scale. Unlike traditional automation tools that require complex integrations, Miniloop combines AI with Python code execution to help you build reliable, repeatable workflows in minutes.

How Miniloop Works

At its core, Miniloop executes workflows - multi-step automations where each step is a Python script that processes data. Here's what makes it powerful:

• AI-Generated Code: Describe what you want to do in plain English, and Miniloop's AI (powered by Google Gemini 2.0 Flash) generates the Python code for you
• Isolated Execution: Each workflow runs in a secure E2B sandbox environment, ensuring your code executes safely without affecting other workflows
• Data Pipeline Architecture: Output from one step automatically feeds into the next, creating seamless data transformations
• Built for Scale: Process hundreds or thousands of files with the same workflow, maintaining consistency and reliability

Common Use Cases

Teams use Miniloop to automate workflows like:

• Data Enrichment: Upload a CSV of company names, automatically fetch website URLs, extract contact information, and export enriched data
• Content Processing: Batch process images to extract text (OCR), analyze sentiment in customer reviews, or generate summaries of long documents
• Data Transformation: Convert between formats (CSV to JSON, XML to CSV), clean messy data, or standardize inconsistent records
• Analysis Automation: Run the same analysis script on multiple datasets, generate reports, or create visualizations

Prerequisites

Before you begin, make sure you have:

1. A Miniloop Account: Sign up at miniloop.ai. The free tier includes everything you need to get started.
2. Sample Data (Optional): For this tutorial, we'll process a CSV file. You can use your own data or download a sample dataset.
3. Basic Understanding of Data: You don't need coding experience, but understanding concepts like rows, columns, and CSV files will help.

No Python knowledge required - Miniloop's AI will generate the code for you!

Step 1: Create Your First Workflow

Let's create a simple workflow that processes a CSV file, analyzes the data, and outputs results.

Access the Workflow Builder

1. Log in to your Miniloop account
2. From your dashboard, click "Create Workflow" or "New Workflow"
3. You'll see the AI workflow builder interface

Describe Your Workflow

Miniloop uses conversational AI to help you build workflows. In the chat interface, describe what you want to accomplish. For this tutorial, try:

"I want to upload a CSV file with customer data, analyze which customers have the highest purchase amounts, and create a new CSV with the top 10 customers."

The AI will ask clarifying questions about your data structure. Answer with details like:

• Column names in your CSV (e.g., "customer_name", "purchase_amount", "email")
• What constitutes "highest" (total amount, average amount, most recent purchase)
• Any filtering criteria (exclude refunds, only include last 90 days, etc.)

Review the Generated Workflow

Based on your description, Miniloop generates a multi-step workflow. A typical workflow for the example above includes:

Step 1: Load CSV Data

• Reads the uploaded CSV file
• Validates data structure
• Outputs a pandas DataFrame

Step 2: Analyze and Filter

• Sorts customers by purchase amount
• Selects top 10 customers
• Formats data for output

Step 3: Export Results

• Converts filtered data to CSV
• Generates downloadable file

Each step shows the Python code that will execute. You can review it to understand what's happening, or trust the AI and proceed.

Step 2: Configure Input Forms

Workflows need data to process. Miniloop automatically creates input forms that let you (or others) submit data when running the workflow.

Understanding Input Types

For our CSV processing example, Miniloop creates a file upload input:

• File Upload: Users can upload CSV files up to a certain size limit
• Text Inputs: For workflow parameters (e.g., "number of top customers to show")
• Dropdowns: For predefined choices (e.g., "sort by: Amount | Date | Frequency")

Customize Your Form

Click on the form preview to customize:

1. Field Labels: Make them descriptive ("Upload Customer Data CSV" instead of "file")
2. Help Text: Add instructions like "CSV must include columns: customer_name, purchase_amount, email"
3. Validation Rules: Require certain fields, set file size limits, or restrict file types

The clearer your form, the easier it is for you (and others) to use your workflow.

Step 3: Test Your Workflow

Before deploying, always test with sample data.

Run a Test Execution

1. Click "Test Run" or "Execute"
2. Fill out your input form with sample data
3. Submit and watch the execution progress

Monitor Execution

Miniloop shows real-time execution status:

• Step Progress: See which step is currently running
• Code Execution: View the Python code as it executes
• Outputs: See intermediate results from each step
• Logs: Check for errors or warnings

For our example workflow, you'll see:

1. Step 1 completes: "Loaded 150 customer records"
2. Step 2 completes: "Filtered to top 10 customers by purchase amount"
3. Step 3 completes: "Exported CSV with 10 records"

Review Results

After execution completes:

• Download output files (the generated CSV with top customers)
• Check execution logs for any warnings
• Verify the results match your expectations

Debugging Failed Executions

If a step fails:

1. Read the Error Message: Miniloop shows Python error messages with line numbers
2. Check Your Data: Verify your CSV has the expected columns and data types
3. Adjust Code: Click on the failed step to modify the Python code
4. Ask the AI: Describe the error in the chat ("The workflow failed with KeyError: 'purchase_amount'") and the AI will help fix it

Common issues:

• Column names don't match (case-sensitive: "Email" vs "email")
• Missing data (null values in required fields)
• Wrong data types (text in a number field)

Step 4: Deploy and Share

Once your workflow works correctly, deploy it for regular use.

Activate Your Workflow

1. Click "Activate" or "Publish"
2. Your workflow moves from "Draft" to "Active" status
3. Active workflows can be run by anyone with the link (if you enable sharing)

Run Your Workflow Regularly

Now you can use your workflow whenever needed:

• Manual Runs: Upload new data through the form and execute
• Scheduled Runs: (Coming soon) Set up automatic execution daily, weekly, or monthly
• API Access: (Coming soon) Trigger executions programmatically via API

Share with Your Team

Miniloop workflows can be shared:

1. View-Only Links: Others can see the workflow structure but not modify it
2. Execution Links: Team members can run the workflow with their own data
3. Team Projects: Organize workflows into projects for better collaboration

Advanced Features

As you become comfortable with basic workflows, explore advanced capabilities:

Multiple File Processing

Process entire datasets at once:

• Upload a ZIP file containing 100 CSV files
• The workflow extracts each CSV, processes it, and combines results
• Perfect for batch data processing

Multi-Step Transformations

Chain complex operations:

1. Extract data from PDF invoices
2. Parse extracted text into structured data
3. Match against existing customer database
4. Generate summary report
5. Email results to stakeholders

Custom Python Libraries

Miniloop supports popular Python libraries:

• pandas: Data manipulation and analysis
• requests: HTTP requests to external APIs
• Pillow (PIL): Image processing
• BeautifulSoup: Web scraping
• OpenAI: Direct LLM API calls for custom AI operations

Import them in your code steps:

import pandas as pd
import requests
from PIL import Image

Error Handling and Retries

Make workflows more robust:

• Add try/catch blocks in Python code
• Implement retry logic for API calls
• Validate data before processing

Common Use Cases in Detail

Use Case 1: Lead Enrichment Workflow

Goal: Take a list of company names, find their websites, extract contact information.

Workflow Steps:

1. Load CSV with company names
2. For each company, search Google for official website
3. Scrape website to find contact page
4. Extract email addresses and phone numbers
5. Export enriched CSV with original data + contact info

Execution Time: ~30 seconds for 50 companies

Use Case 2: Image Text Extraction (OCR)

Goal: Upload photos of receipts, extract text, categorize expenses.

Workflow Steps:

1. Load uploaded images
2. Run OCR (Optical Character Recognition) on each image
3. Parse extracted text for date, vendor, amount, items
4. Categorize expenses (office supplies, travel, meals, etc.)
5. Generate expense report CSV

Execution Time: ~2 seconds per image

Use Case 3: Sentiment Analysis on Reviews

Goal: Analyze customer reviews to identify common complaints.

Workflow Steps:

1. Load CSV of customer reviews (text + rating)
2. Run sentiment analysis on each review using AI
3. Extract key phrases and topics
4. Group reviews by sentiment (positive, neutral, negative)
5. Generate summary report with most common issues

Execution Time: ~5 seconds for 100 reviews

Best Practices

Follow these guidelines for reliable workflows:

1. Start Simple

Begin with a basic workflow (input → process → output) before adding complexity. Test thoroughly at each stage.

2. Validate Your Data

Always check data quality:

• Verify column names and data types
• Handle missing values explicitly
• Test with edge cases (empty files, single row, very large files)

3. Add Descriptive Names

Name your workflows clearly:

• ✅ "Customer Purchase Analysis - Top 10"
• ❌ "Workflow 1"

4. Document Your Workflows

Add comments in Python code explaining complex logic. Future you (and teammates) will thank you.

5. Version Your Workflows

When making major changes:

1. Duplicate the existing workflow
2. Modify the copy
3. Test thoroughly
4. Only then replace the original

This gives you a rollback option if something breaks.

Troubleshooting

Workflow Won't Execute

Check:

• Is your workflow activated?
• Did you fill out all required form fields?
• Are you within your account's execution limits?

Results Don't Match Expectations

Check:

• Review the code generated by AI
• Verify your input data format
• Look at execution logs for warnings
• Run a test with a small sample dataset first

Step Fails Halfway Through

Check:

• Does the error message indicate missing data?
• Did you import required Python libraries?
• Are you trying to access columns that don't exist?

What's Next?

Now that you've created your first workflow, here are ways to expand your automation capabilities:

Explore Templates

Miniloop provides pre-built workflow templates:

• CSV Transformations: Merge, split, filter, and clean datasets
• Data Enrichment: Add external data from APIs
• Report Generation: Create formatted reports from raw data
• Image Processing: Batch resize, convert, or analyze images

Browse templates in the dashboard and customize them for your needs.

Join the Community

Connect with other Miniloop users:

• Discord Community: Ask questions, share workflows, get help
• Blog & Tutorials: Learn advanced techniques and best practices
• Feature Requests: Suggest improvements to the platform

Upgrade Your Plan

As your automation needs grow:

• More Executions: Process larger datasets or run workflows more frequently
• Longer Execution Time: Handle complex workflows that take minutes to complete
• Team Collaboration: Add team members and share workflows across your organization
• Priority Support: Get help faster when you need it

Conclusion

Congratulations! You've learned how to create, test, and deploy AI-powered workflows in Miniloop. You now have the foundation to automate repetitive data processing tasks, saving hours of manual work.

Remember:

• Start with simple workflows and gradually increase complexity
• Test thoroughly with sample data before processing production datasets
• Use the AI to generate code, but review it to understand what's happening
• Document and version your workflows for maintainability

Ready to build your next workflow? Head to the Miniloop dashboard and start automating!

Additional Resources

• Documentation: docs.miniloop.ai - Comprehensive guides and API reference
• Example Workflows: miniloop.ai/templates - Pre-built workflows you can copy
• Video Tutorials: YouTube Channel - Step-by-step video guides
• Support: support@miniloop.ai - Get help from the team

Happy automating!