All services are Cloud-hosted. Yet we offer On-Prem Hosting for data-critic Enterprises.

​
The Promise

Integrate in minutes our privacy-first reverse proxy into your running applications. This will automatically obfuscate PIIs before they’re sent to 3rd party softwares eg. LLMs, DeepL, Email Clients and much more. Dealing with sensitive data in your pipelines is not a problem anymore!

​
How It Works

Ultimately, it is our REST API that will do the heavy-lifting. That is detecting, masking and clearing PIIs from your text (or other file formats). Yet we provide fully open-source Bindings (SKDs) to common modern programming language frameworks.

This way you can directly import Filtro into your codebase and get access to top-level data privacy in only two single lines of code. We currently have bindings for Python (pip), Nodejs (npm) and Rust (Cargo) for now. We will keep adding more of them on the way.

Shoot us an email and we’ll integrate the Binding you need asap, at no extra charge!

​
Bindings (SDKs)

Python Binding

See how to integrate Filtro to your Python scripts, Langchain and more

Nodejs SDK

See how to integrate Filtro to your Nodejs powered web applications

Rust SDK

Power your browser with transparent LLM’s privacy-layer extensions now

REST API

Explore the source documentation and settings of our internal REST API

​
Prerequisites

You will need an API key to connect to our REST API or equivalent SDKs.

Visit our Platform to request yours. Learn more about Authentication.

​
An example

Below is a simple example adapted from Langchain, to extract information from unstructured text and parse it into a well-defined structure. You can observe that with only 3 lines of added code (highlighted), you are able to mask-out any sensitive data.

from langchain.chat_models import ChatOpenAI
from langchain.chains import create_extraction_chain
from filtro import mask, clear # Filtro SDK

# Define Schema (LangChain Codebase)
schema = {
    "properties": {
        "name": {"type": "string"},
        "height": {"type": "integer"},
        "hair_color": {"type": "string"},
    },
    "required": ["name", "height"],
}

# Original Input (LangChain Codebase)
inp = """
Alex is 5 feet tall.
Claudia is 1 feet taller Alex and jumps higher than him.
Claudia is a brunette and Alex is blonde.
"""

masked, context = mask(inp) # Filtro Masking

# Run chain (LangChain Codebase)
llm = ChatOpenAI(temperature=0, model="gpt-3.5-turbo")
chain = create_extraction_chain(schema, llm)
llm_out = chain.run(masked) # Call Chain with `masked` data

clear(llm_out) # Filtro Clearing

Of course, you can scale this to much larger applications and intricate systems.

FAQBrowser Extension