Skip to content
Integradio

Getting Started

This guide walks you through installing Integradio, configuring Ollama for local embeddings, and building your first semantically-searchable Gradio app.

Prerequisites

Section titled “Prerequisites”
  • Python 3.10+
  • Gradio 4.0+ (compatible with Gradio 5.x and 6.x)
  • Ollama installed and running locally

Install Integradio

Section titled “Install Integradio”
Terminal window
# Basic installation
pip install integradio


# With all optional dependencies (visualization, FastAPI routes, etc.)
pip install "integradio[all]"


# Development installation (from source)
pip install -e ".[dev]"

Set up Ollama

Section titled “Set up Ollama”

Integradio uses Ollama to generate vector embeddings locally. No cloud APIs required.

Terminal window
# Install Ollama from https://ollama.ai/
# Then pull the embedding model:
ollama pull nomic-embed-text


# Start the Ollama server (if not already running):
ollama serve

The default embedding model is nomic-embed-text. It runs on your GPU and produces 768-dimensional vectors suitable for semantic search.

Wrap your first component

Section titled “Wrap your first component”

The semantic() function wraps any Gradio component with an intent string. This intent gets embedded into a vector, making the component discoverable by meaning.

import gradio as gr
from integradio import SemanticBlocks, semantic


with SemanticBlocks() as demo:
    # Wrap a Textbox with semantic intent
    query = semantic(
        gr.Textbox(label="Search Query"),
        intent="user enters search terms"
    )


    # Wrap a Button
    search_btn = semantic(
        gr.Button("Search"),
        intent="triggers the search operation"
    )


    # Wrap an output area
    results = semantic(
        gr.Markdown(),
        intent="displays search results"
    )


    # Wire up events as usual
    search_btn.click(fn=lambda q: f"Results for: {q}", inputs=query, outputs=results)


demo.launch()

Search by meaning

Section titled “Search by meaning”

Once components are wrapped, you can search for them semantically:

# Find all components related to "user input"
matches = demo.search("user input", k=5)
for match in matches:
    print(f"{match.label} (score: {match.score:.3f})")


# Get the single best match
best = demo.find("where does the user type?")
print(best.label)  # "Search Query"


# Print a summary of all registered components
print(demo.summary())

Configuration options

Section titled “Configuration options”

SemanticBlocks accepts several configuration parameters:

with SemanticBlocks(
    db_path=None,           # SQLite path for persistent storage (None = in-memory)
    cache_dir=None,         # Directory for embedding cache
    ollama_url="http://localhost:11434",  # Ollama server URL
    embed_model="nomic-embed-text",      # Embedding model name
) as demo:
    ...

Next steps

Section titled “Next steps”