DecisionUI

Turn LLM API into Decision Model API.

DecisionUI is a lightweight Python server that sits in front of a Chat Completions endpoint (vLLM, SGLang, Ollama, a hosted API, etc.) and exposes it through two decision APIs:

• SystemOne API
• OpenAI Decisions API

Instead of parsing free-form text, you ask structured questions and get back the chosen option along with its probabilities.

Installation

pip install decision-ui

Quick Start

1. Serve a model with any OpenAI-compatible inference engine, for example vLLM:

vllm serve google/gemma-4-31B-it --host localhost --port 8000

2. Launch DecisionUI and point it at the model server:

decision-ui serve http://localhost:8000 --host localhost --port 9000

3. Use a LLM as a Decision Model

from typesafe_sdk import Choice, TypeSafeClient

with TypeSafeClient(base_url="http://localhost:9000", api_key="EMPTY") as client:
    decision = client.system_one(
        model="google/gemma-4-31B-it",
        state={"document": "I was charged twice. Please fix this ASAP."},
        questions={
            "category": Choice(
                instructions="What is this ticket about?",
                criteria={"billing": None, "technical": None, "other": None},
            ),
        },
    )

print(decision.choices["category"].probabilities)

Example output:

{"billing": 0.94, "technical": 0.03, "other": 0.03}

How it works

Your app ──► DecisionUI (:9000) ──► Chat Completions server (:8000) ──► LLM
          SystemOne / Decisions API        e.g. vLLM

CLI Reference

decision-ui serve <backend_url> [--host HOST] [--port PORT]