Where Does Minusx Keep My Data and How Do I Run It?

Updated

Minusx gives you two deployment paths, and the choice determines where your data lives. If you self-host the open-source version, your warehouse stays in your own environment. If you use MinusX Cloud, you get a managed runtime for faster setup, with the option to bring your own LLM keys. The install path is a single shell script or a clone from GitHub.

Self-host: your data stays in your environment

The open-source route is built for teams that need their warehouse to remain inside their own infrastructure. Minusx describes the core as source-available and inspectable, so you can review what runs before you point it at production data.

What this means in practice:

  • Your warehouse stays in your env. Minusx connects to your data where it already lives rather than copying it into a vendor system.
  • You control the runtime. The agent runs in your own environment, so network boundaries and access rules are yours to set.
  • The core is inspectable. You can read the source rather than trusting a black box.

This is the right choice when data residency, internal security review, or existing warehouse governance rules make a hosted tool a non-starter.

Cloud: managed runtime, optional BYO keys

MinusX Cloud trades some control for speed. The runtime is managed for you, which removes the setup work of standing up and maintaining the agent yourself.

Two things worth noting:

  • Fast setup is the point. The site positions Cloud as the option to use "when speed matters."
  • You can bring your own LLM keys. If your organization already has LLM provider agreements or wants inference to run under its own account, Cloud supports BYO keys.

Cloud is the better fit when you want to evaluate the product quickly, or when you don't have the infrastructure appetite to run and maintain the agent internally.

How to get started

Minusx documents two entry points:

Install script (self-host or cloud mode):

curl -fsSL minusx.ai/install.sh | bash

The installer presents a mode choice — self-host or cloud — and a keys choice — yours or managed. That combination is the whole deployment decision in one step.

From source: the GitHub repository is linked from the site for teams that want to build or audit the code directly.

Try the hosted version: the site offers a free trial of MinusX Cloud, and a "Book a Demo" option if you want a guided walkthrough first.

Choosing between them

Self-host (OSS) MinusX Cloud
Where data lives Your environment / your warehouse Managed runtime
Setup effort You run and maintain it Fast, managed
LLM keys Yours Managed, or BYO keys
Source access Source-available, inspectable Not applicable
Best for Data residency and security requirements Quick evaluation and low ops overhead

If your constraint is "the data cannot leave our environment," self-host is the only option that satisfies it. If your constraint is "we need to see this working this week," Cloud gets you there faster, and BYO keys let you keep inference under your own account.

A practical starting sequence

  1. Decide the constraint first: does your data have to stay in your own environment? If yes, self-host. If no, Cloud is the faster path.
  2. For self-host, run the install script, choose self-host, and confirm the agent can reach your warehouse before pointing it at anything sensitive.
  3. For Cloud, start the free trial and connect a non-critical data source first — a Google Sheets export or a staging warehouse — so you can judge answer quality before widening access.
  4. Either way, test the agent against questions whose answers you already know. Minusx ships an Evals feature for exactly this: checking whether the agent understands your company's definitions, such as what counts as ARR or an active user.

That last step matters more than the deployment choice. A self-hosted agent that misreads your metric definitions is no more trustworthy than a hosted one — the deployment decision controls where data lives, not whether the answers are right.