Introduction to Gateway

Gateway fact-checks LLM-generated text against your source documents. You provide a claim and sources, and Gateway tells you whether the claim is supported, refuted, or lacks sufficient evidence.

AWS
Azure

from gateway import TextSource
from gateway.aws import AwsClient

client = AwsClient.from_sfn_arn("arn:aws:states:us-east-1:123456789:stateMachine:gateway")

ruling = client.judge(
    claim="The Eiffel Tower is in Paris and was built in 1889.",
    sources=[
        TextSource(id="doc1", text="The Eiffel Tower is located in Paris, France."),
        TextSource(id="doc2", text="It was completed in 1889."),
    ],
)

print(ruling.verdict)  # Verdict.SUPPORTS

import asyncio
from gateway import TextSource
from gateway.azure import AzureClient

client = AzureClient.from_function_app_url(
    "https://yourprefix-func.azurewebsites.net",
    api_key="your-function-key",
)

ruling = asyncio.run(client.judge(
    claim="The Eiffel Tower is in Paris and was built in 1889.",
    sources=[
        TextSource(id="doc1", text="The Eiffel Tower is located in Paris, France."),
        TextSource(id="doc2", text="It was completed in 1889."),
    ],
))

print(ruling.verdict)  # Verdict.SUPPORTS

How it works

Claim decomposition - Gateway breaks your claim into individual statements
Evidence retrieval - Each statement is matched against relevant passages in your sources
Verdict classification - We determine whether each statement is supported, refuted, or lacks evidence

Supported clouds

Gateway deploys to your cloud account. We currently support AWS and Azure. During onboarding, we provide the SDK and Terraform configuration for your cloud via SFTP.

How it works

Supported clouds

Next steps

Deploy on AWS

Deploy on Azure

Install the SDK

​How it works

​Supported clouds

​Next steps

Deploy on AWS

Deploy on Azure

Install the SDK

How it works

Supported clouds

Next steps