Quickstart - Structify

Our API lets you transform any information - from documents to web pages - into structured data with just a few API calls.

Installation

First, install the Python client library:

pip install structifyai

uv pip install structifyai

poetry add structifyai

We constantly push updates. Keep your client current with pip install structifyai --upgrade

Authentication

Get your API key from the Structify Dashboard and set it as an environment variable:

export STRUCTIFY_API_TOKEN="your_api_key"

Then import and initialize the client:

from structify import Structify

# Uses STRUCTIFY_API_TOKEN env var automatically
client = Structify()

# Or provide explicitly
client = Structify(api_key="your_api_key")

Create Your First Dataset

Let’s build a simple book and author dataset in three steps:

1. Define Your Schema

Create tables for authors and books, plus a relationship between them:

from structify import Structify
from structify.types.table import Table, Property
from structify.types.dataset_descriptor import Relationship

client = Structify()

# Define entity tables
tables = [
    Table(
        name="author",
        description="an individual who wrote a book",
        properties=[
            Property(name="name", description="The name of the author"),
            Property(name="genre", description="The genre the author typically writes in"),
            Property(name="birth_year", description="Year the author was born", prop_type="Integer")
        ]
    ),
    Table(
        name="book",
        description="a book that has been written",
        properties=[
            Property(name="title", description="The title of the book"),
            Property(name="publication_year", description="Year published", prop_type="Integer"),
            Property(name="copies_sold", description="Number of copies sold", prop_type="Integer")
        ]
    )
]

# Define relationships
relationships = [
    Relationship(
        name="authored_by",
        description="Connects a book to its author",
        source_table="book",
        target_table="author"
    )
]

# Create the dataset
client.datasets.create(
    name="books",
    description="Books and their authors",
    tables=tables,
    relationships=relationships
)

2. Add Initial Data

Add some authors to start with:

from structify.types import KnowledgeGraphParam, EntityParam

authors = ["J.K. Rowling", "Stephen King", "Harper Lee"]

for author_name in authors:
    client.entities.add(
        dataset="books",
        kg=KnowledgeGraphParam(
            entities=[
                EntityParam(
                    id=0,
                    type="author",
                    properties={"name": author_name}
                )
            ]
        )
    )

3. Enrich Your Dataset

Use Structify’s AI to find and add information:

# For each author, find their genre and books
for author in client.datasets.view_table(dataset="books", table="author"):
    # Fill in missing properties
    client.structure.enhance_property(
        entity_id=author.id,
        property_name="genre"
    )

    client.structure.enhance_property(
        entity_id=author.id,
        property_name="birth_year"
    )

    # Find books they've written
    client.structure.enhance_relationship(
        entity_id=author.id,
        relationship_name="authored_by"
    )

Complete Example

Here’s everything together in one script:

from structify import Structify
from structify.types import KnowledgeGraphParam, EntityParam
from structify.types.table import Table, Property
from structify.types.dataset_descriptor import Relationship

def create_book_dataset():
    client = Structify()

    # Step 1: Define schema
    tables = [
        Table(
            name="author",
            description="an individual who wrote a book",
            properties=[
                Property(name="name", description="The name of the author"),
                Property(name="genre", description="The genre the author typically writes in"),
                Property(name="birth_year", description="Year born", prop_type="Integer")
            ]
        ),
        Table(
            name="book",
            description="a book that has been written",
            properties=[
                Property(name="title", description="The title of the book"),
                Property(name="publication_year", description="Year published", prop_type="Integer"),
                Property(name="copies_sold", description="Copies sold", prop_type="Integer")
            ]
        )
    ]

    relationships = [
        Relationship(
            name="authored_by",
            description="Connects a book to its author",
            source_table="book",
            target_table="author"
        )
    ]

    # Create dataset
    client.datasets.create(
        name="books",
        description="Books and their authors",
        tables=tables,
        relationships=relationships
    )

    # Step 2: Add authors
    for author in ["J.K. Rowling", "Stephen King", "Harper Lee"]:
        client.entities.add(
            dataset="books",
            kg=KnowledgeGraphParam(
                entities=[EntityParam(id=0, type="author", properties={"name": author})]
            )
        )

    # Step 3: Enrich with AI
    for author in client.datasets.view_table(dataset="books", table="author"):
        client.structure.enhance_property(entity_id=author.id, property_name="genre")
        client.structure.enhance_relationship(entity_id=author.id, relationship_name="authored_by")

    print("Dataset created and enriched!")

    # View results
    results = client.datasets.view_table(dataset="books", table="author", limit=10)
    for author in results:
        print(f"Author: {author.properties['name']}, Genre: {author.properties.get('genre', 'Unknown')}")

if __name__ == "__main__":
    create_book_dataset()

What’s Next?

Now that you’ve created your first dataset, you can:

Explore Schema Design

Learn best practices for designing schemas

Process Documents

Extract data from PDFs and documents

Web Scraping

Pull structured data from websites

View Examples

See real-world use cases

Key Capabilities

After reading our documentation, you’ll be able to:

Create personalized datasets of your professional network
Monitor changes in continuously updating datasets
Extract structured data from SEC filings and pitch decks
Automate notifications when new entities match your criteria
Build structured datasets from any unstructured data source

​Installation

​Authentication

​Create Your First Dataset

​1. Define Your Schema

​2. Add Initial Data

​3. Enrich Your Dataset

​Complete Example

​What’s Next?

Explore Schema Design

Process Documents

Web Scraping

View Examples

​Key Capabilities

Installation

Authentication

Create Your First Dataset

1. Define Your Schema

2. Add Initial Data

3. Enrich Your Dataset

Complete Example

What’s Next?

Key Capabilities