---
title: "OU.Chat: Enterprise AI that never leaves the building. White-label, on your own models."
description: "OU.Chat is a white-label AI app enterprises run on their own models and data. We built it so a regulated organization can give staff and customers real AI support without a single prompt reaching OpenAI or Google. The intelligence comes to the data, not the other way around."
source: "https://www.kensink.com/cases/ou-chat/"
---
ENTERPRISE AI · SOVEREIGN INFRASTRUCTURE OU.Chat

# OU.Chat.  
Enterprise AI that never leaves the building.

White-label, on your own models.

0

Prompts sent to public models

WHITE-LABEL PRIVATE AI ENTERPRISE AI · SOVEREIGN INFRASTRUCTURE

## OU.Chat: Enterprise AI that never leaves the building. White-label, on your own models.

OU.Chat is a white-label AI app enterprises run on their own models and data. We built it so a regulated organization can give staff and customers real AI support without a single prompt reaching OpenAI or Google. The intelligence comes to the data, not the other way around.

[View .md](https://www.kensink.com/cases/ou-chat.md)

0

Prompts sent to public models

White-label

Branded for staff and customers

In-boundary

Models run inside the customer VPC

Owned

Tools persist past the contract

01 · THE PROBLEM

## Where they were stuck.

Enterprises with a data-protection duty were stuck. Public AI meant shipping sensitive records to a third-party model, which legal would not allow. So they either banned AI outright or tolerated shadow usage they could not see. The brief was to make private, sovereign AI feel as easy as the public tools staff already wanted to use.

02 · OUR APPROACH

## How we built it.

-   01 White-label mobile app: one codebase an enterprise brands as its own and ships to staff and customers, each on a separate permissioned surface
-   02 Self-hosted, open-weight models inside the customer's boundary, behind a vendor-neutral abstraction so the model can be swapped without touching the app
-   03 Private retrieval over the organization's own documents and systems, so answers come from what the company knows, for internal and external use
-   04 A control plane for roles, guardrails, tone, and access, so the enterprise decides exactly what each audience can do
-   05 Tools and knowledge stored in the customer's own instance, so what they build persists past any subscription

> “Sovereign AI usually means a worse product. The point of OU.Chat was to make the private option the good option, so nobody is tempted to paste company data into a public chatbot to get a better answer.”

_Lab engineering

Kensink Labs_

\[TECH STACK\]

-   Self-hosted LLMs
-   vLLM
-   React Native
-   Private RAG
-   Role-based access

\[ENGAGEMENT\]

Duration Ongoing (product)

Client OU.Chat

Shape WHITE-LABEL PRIVATE AI

Handoff Full ownership · 90-day warranty

START YOUR OWN PROJECT

## Bring a real problem.  
We’ll bring code on day one.

[Start your own project →](https://www.kensink.com/contact) [Book a 15-min intro](https://www.kensink.com/contact)

[← All cases](https://www.kensink.com/cases) CASE · OU-CHAT