US Patent 12,413,397 · Open-source SDK

Consent infrastructure for AI API calls

Prove you honoured your users' choices. ID Side wraps your AI provider client, enforces each user's consent on every request, and hands you a signed, independently verifiable proof that it happened.

Create your account How it works

from idside import consent
from openai import OpenAI

client = consent.wrap(OpenAI())

client.chat.completions.create(
    model="gpt-4o",
    messages=[...],
    user_id="usr_marie_456",
)
# Consent resolved, parameters forced,
# ACT token signed, ledger entry appended.

What happens on every call

01

Consent resolved
02

Parameters forced
03

ACT token signed
04

Ledger entry appended

US 12,413,397 · EP4100905

Patents granted in the United States and the European Union
Apache-2.0

The Python and Node SDKs, open source
1 call = 1 token

Bound to that one request, and verifiable without us

SDK Overview

What is in here

Three pages and one library. Read them in this order and you will know what the SDK enforces, what it never sees, and how to have it running against your own provider account.

How it works

The two circuits — a preference resolved before every call, and the signed proof that comes back — and the second source of a preference: an organization setting a policy for the people who work under it.

Security model

What ID Side never receives, what the ledger proves, and where enforcement stops. Each provider carries an honest, sourced trust level rather than a claim we could not keep.

Quickstart

Install the SDK, point it at your account, wrap your provider client, and verify the token that comes back. Four steps, and the published API contract underneath them.

Reference documentation

The SDK's own documentation set: concepts, errors, provider coverage, and a quickstart for Python and for Node. It lives with the source, in the open-source repository.