Quickstart

Create your first AI-powered endpoint in just a few steps.

This guide assumes you have already installed Syft Space. If not, install it first using Docker, Desktop App, or from source.

Step 1: Register your account

After installation, open Syft Space in your browser:

http://localhost:8080

Click Register

Create your account with email and password

Configure network

Local deployment
Cloud/VM deployment

Add your developer token (contact OpenMined to get one)

Developer tokens are only needed for local deployments when publishing endpoints. See Installation for details.

Step 2: Add your data

Create a dataset to store your documents:

Navigate to Datasets

Click Datasets in the sidebar

Click Add Dataset

Fill in the dataset details:

Name: my-docs
Type: Local file (ChromaDB)
Summary: My first dataset

Upload files

Upload PDF, TXT, or DOCX files. They’ll be automatically:

Extracted and chunked
Embedded with OpenAI
Indexed in ChromaDB

Wait for processing

Monitor ingestion progress in the dataset detail view

Start with a few small documents to test. You can add more later.

Step 3: Connect a model

Configure an AI model to answer queries:

Navigate to Models

Click Models in the sidebar

Click Add Model

Fill in the model details:

Name: gpt-4
Type: OpenAI
API Key: Your OpenAI API key
Model: gpt-4 (or gpt-3.5-turbo for lower cost)

Test connection

The health check will verify your API key is valid

Never commit your API keys to version control. Store them securely in environment variables.

Step 4: Create an endpoint

Combine your dataset and model into a queryable endpoint:

Navigate to Endpoints

Click Endpoints in the sidebar

Click Add Endpoint

Fill in the endpoint details:

Name: My Docs Assistant
Slug: my-docs (must be unique)
Dataset: my-docs
Model: gpt-4
Response Type: Both (summary + references)

Configure policies (optional)

Add access control or rate limiting policies

Save endpoint

Your endpoint is now ready to query!

Step 5: Test your endpoint

Query your endpoint using the built-in interface or API:

Web interface
cURL
Python

Open your endpoint detail page
Use the built-in query interface
Ask a question: “What are the main topics in these documents?”
View the AI-generated summary and source references

curl -X POST http://localhost:8080/api/v1/endpoints/my-docs/query \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -d '{
    "messages": [
      {"role": "user", "content": "What are the main topics?"}
    ]
  }'

import requests

response = requests.post(
    "http://localhost:8080/api/v1/endpoints/my-docs/query",
    headers={
        "Content-Type": "application/json",
        "Authorization": "Bearer YOUR_TOKEN"
    },
    json={
        "messages": [
            {"role": "user", "content": "What are the main topics?"}
        ]
    }
)

result = response.json()
print(result["summary"]["message"]["content"])

Step 6: Publish (optional)

Make your endpoint discoverable on SyftHub:

Click Publish

In the endpoint detail page, click the Publish button

Choose visibility

Organization: Only accessible to your team
Public: Discoverable on SyftHub marketplace

Confirm

Your endpoint is now live at syfthub.openmined.org

Publishing requires either a developer token (local) or public IP (cloud). Configure this in Settings → Network.

What’s next?

Core concepts

Understand how datasets, models, endpoints, and policies work

Creating datasets

Learn about dataset types and file ingestion

Building endpoints

Master endpoint configuration and testing

API reference

Explore the complete REST API

Example use case

Company knowledge base:

Dataset: Upload your internal documentation (wikis, guides, policies)
Model: Connect GPT-4 for high-quality answers
Endpoint: Create “company-kb” endpoint with summary + references
Policies:
- Access: Only allow @company.com emails
- Rate limit: 100 requests/hour per user
Publish: Share within organization

Now employees can ask questions like “What is our vacation policy?” and get accurate answers with source citations.

Start simple and iterate. You can always add more datasets, refine policies, and optimize performance later.

Get Started

Core Concepts

Guides

Desktop App

Deployment

Advanced

Step 1: Register your account

Step 2: Add your data

Step 3: Connect a model

Step 4: Create an endpoint

Step 5: Test your endpoint

Step 6: Publish (optional)

What’s next?

Core concepts

Creating datasets

Building endpoints

API reference

Example use case

Build docs developers (and LLMs) love

Get Started

Core Concepts

Guides

Desktop App

Deployment

Advanced

​Step 1: Register your account

​Step 2: Add your data

​Step 3: Connect a model

​Step 4: Create an endpoint

​Step 5: Test your endpoint

​Step 6: Publish (optional)

​What’s next?

Core concepts

Creating datasets

Building endpoints

API reference

​Example use case

Build docs developers (and LLMs) love

Step 1: Register your account

Step 2: Add your data

Step 3: Connect a model

Step 4: Create an endpoint

Step 5: Test your endpoint

Step 6: Publish (optional)

What’s next?

Example use case