Extract Knowledge Graphs in 5 Minutes

Turning unstructured documents into structured, queryable knowledge is painfully slow. You either spend hours manually tagging entities and relationships, or you build complex extraction pipelines that break on edge cases. Graphora solves this with AI-powered knowledge graph extraction that works out of the box -- no schema definition, no configuration, no database setup required.

Option 1: Try the Demo (30 seconds)

The fastest way to see Graphora in action:

1. Visit demo.graphora.io
2. Upload a PDF or paste text directly
3. Watch the AI automatically extract entities and relationships
4. Explore the interactive graph visualization

No signup, no credit card, no installation. Just drag, drop, and explore.

Option 2: Google Colab (2 minutes)

Run Graphora in your browser with zero setup:

The notebook walks you through:

• Installing Graphora with one line
• Configuring your Gemini API key
• Extracting a knowledge graph from your documents
• Exploring and visualizing the results

Perfect for experimentation and learning the API without touching your local environment.

Option 3: CLI (5 minutes)

For developers who prefer the command line:

1# Install the CLI
2pip install graphora[cli]
3
4# Configure your Gemini API key (get one free at https://aistudio.google.com/apikey)
5graphora config init --api-key "YOUR_GEMINI_API_KEY"
6
7# Extract a knowledge graph
8graphora extract document.pdf --output graph.json
9
10# Or extract directly to Cypher format for Neo4j
11graphora extract document.pdf --format cypher --output import.cypher

That's it. Graphora runs in-memory by default and infers the schema automatically from your documents. The only requirement is a Gemini API key for the AI extraction.

Using the Python Client

For programmatic access, install the Python client:

1pip install graphora

Then use it in your code:

1from graphora import GraphoraClient
2
3client = GraphoraClient(base_url="http://localhost:8000")
4
5# Upload and extract
6result = client.extract("report.pdf")
7
8# Query entities
9entities = result.entities
10relationships = result.relationships

What's Happening Under the Hood?

When you run graphora extract, here's what happens:

1. Document Parsing: Extracts text from PDFs, DOCX, HTML, or plain text files
2. Schema Inference: AI analyzes the content and determines relevant entity types and relationship patterns
3. Entity Extraction: Identifies and extracts entities (people, organizations, concepts, dates, etc.)
4. Relationship Detection: Maps connections between entities based on context
5. Deduplication: Merges duplicate entities using semantic similarity and fuzzy matching

All of this runs automatically. You get a clean, queryable knowledge graph without writing a single line of schema code.

Why Graphora?

Customizing the Schema

Want more control? Define your own schema in YAML:

1# schema.yaml
2entities:
3  - name: Person
4    properties:
5      - name
6      - title
7      - email
8  - name: Company
9    properties:
10      - name
11      - industry
12      - location
13
14relationships:
15  - type: WORKS_AT
16    source: Person
17    target: Company
18  - type: PARTNERS_WITH
19    source: Company
20    target: Company

Then pass it to the CLI:

1graphora extract report.pdf --schema schema.yaml --output graph.json

Graphora will extract only the entities and relationships you've defined, giving you full control over the output structure.

Next Steps

Ready to go deeper? Here's where to go next:

• Interactive Schema Designer: Build custom schemas visually with the drag-and-drop ontology builder
• API Documentation: Integrate Graphora into your applications with the REST API or Python SDK
• Self-Hosting Guide: Deploy Graphora in your infrastructure with full control and privacy
• GitHub Repository: Browse the source code, report issues, and contribute

Need Help?

• GitHub Discussions: Ask questions, share use cases, and connect with the community
• Email: [email protected]

Start extracting knowledge graphs today. No PhD in NLP required.