⚙️ What Is a Custom Extractor?

A Custom Extractor lets you define exactly what data to extract from documents or text. You decide the structure, fields, and rules—UpBrains AI does the rest.

Use it when:

·        Prebuilt extractors don’t fit your unique document format

·        You want full control over the fields and output

·        You need to extract structured or tabular data (like line items)

🧠 How to Create a Custom Extractor in UpBrains

Follow this step-by-step guide to build your own extractor:

1️⃣ Navigate to the Extractor Section

Open the UpBrains platform and go to the Extractor section in the left-hand sidebar.
Click Create Custom Extractor.

2️⃣ Step 1: Basic Information

·        Name: Choose a clear and descriptive name (e.g., “PO Extractor” or “Support Email Parser”).

·        Document Class (Optional): Select a document category like Invoice, Email, Quote, etc.

·        Instructions: Help the AI understand what to extract.
Example: “Extract PO number, buyer name, list of items, and total amount.”

3️⃣ Step 2: Define Line Item Keys (if your doc includes tables)

For documents with repeating data (e.g., itemized tables), define Line Item Keys, such as:

·        Product Name

·        Quantity

·        Unit Price

·        Total

4️⃣ Step 3: Define Line Items

Add all the fields you want to extract from the document, such as:

·        Invoice Number

·        Supplier Name

·        Invoice Date

·        Total Amount

·        Payment Terms

·        Any custom metadata

You can combine both flat fields and tabular ones.

5️⃣ Step 4: Test Your Extractor

Upload a sample document—PDF, image, or text—or paste raw content (like an email body).
Click Extract to preview what the AI pulls.

6️⃣ Step 5: Refine the Results

Not getting the right fields? You can:

·        Edit field definitions

·        Clarify extraction instructions

·        Add more example files

This iterative testing ensures high precision.

7️⃣ Step 6: Publish Your Extractor

Once you're happy with the output, click Publish.
Your extractor now becomes usable inside any AI Agent.

💡 Each extractor has a Draft version for editing and a Published version used in production.

🔄 What Happens Next?

Once published, your custom extractor can be used in any workflow—like processing invoices, extracting customer data from forms, or parsing orders from emails. Just plug it into the “Extract Information” step inside your automation flows.

🚀 Why It Matters

·        ✅ Precision: You decide what gets extracted

·        📄 Versatility: Works across PDFs, scanned images, and plain text

·        🤖 Automation-Ready: Connect directly to AI Agents

·        📈 Scalable: Extract thousands of docs with consistent accuracy

🌟 Real-World Use Cases

·        Finance Teams: Automate invoice and PO data capture

·        Operations: Extract structured info from logistics or shipping docs

·        Sales: Parse client details and product requests from emails

·        Customer Support: Structure incoming tickets or inquiries

Start Building Smarter

Let UpBrains do the heavy lifting—just define what matters.
Book a Demo