Valet

Help

Setup

1. Install

Download the installer from the Download page (a free account is required) and run it. Installing needs administrative rights on your PC. The installer asks which model provider and which web-search provider you plan to use; you can change either later.

2. Get an API key, and add it to Valet

Valet uses an AI model from a provider you choose, and you connect it with your own API key. If you have never created one, this is the step to take slowly. It takes about as long as signing up for any online account.

Choose a model provider

Valet comes ready for Anthropic (Claude models), OpenAI (GPT models), Google (Gemini models), Kimi, DeepSeek and Qwen. Anthropic is the installer's default. Any other provider with an OpenAI-compatible API can be added in Valet's settings.

Create the key on the provider's website

  1. Sign up or sign in on the provider's developer site - the Anthropic Console, the OpenAI Platform, or Google AI Studio, for the three most common.
  2. Open its billing page and add a payment method or buy prepaid credit. API use is billed separately from any chat subscription you may already have.
  3. Open its API keys page and create a new key. Give it a name you'll recognise, such as "Valet".
  4. Copy the key straight away. Most providers show it only once.

What a key looks like

A long string of letters, numbers and dashes, usually with a prefix that names the provider: Anthropic keys start sk-ant-, OpenAI keys start sk-, and Google keys start AIza. Treat it like a password: anyone who has it can spend from your account.

Where to paste it

In Valet, open Settings, then the AI Models tab. Choose your provider, paste the key into its API Key field, and save. Then pick which model Valet should use.

The key stays on your PC. This website has no field for one, and you should never paste a key into a website form, an email or a feedback message.

Web search is set up the same way, on the Web Search tab. Brave is the default and needs a key from the Brave Search API. Tavily, Exa, Perplexity, Firecrawl, Gemini, Grok, MiniMax, SearXNG and Ollama are also supported. DuckDuckGo needs no key at all.

What it costs

You pay the provider directly, for what you use; Noerion Labs charges nothing. Providers charge by the amount of text a model reads and writes, so a month's cost depends on which model you choose and how much you ask of it - smaller, faster models cost a fraction of the largest. Each provider publishes its prices on its own site.

To stay in control, set a monthly spending limit in the provider's billing settings, or buy a fixed amount of prepaid credit: when it runs out, requests stop until you top up.

3. First run

Start Valet, add your key as above if you haven't yet, and give it a first task.

4. Connect your accounts

To work with your Google or Microsoft email, calendar and files, sign in to that account from the Integrations tab in Valet's settings. You can see and remove each connection there.

Frequently asked questions

Why do I need an account to download?

An account lets us tie a download to a person, which is how release distribution and support are managed. It's free to create.

How do I get a model-provider API key, and what will it cost?

Create one on your provider's developer site and paste it into Valet's settings - the step-by-step guide above covers it. You pay the provider for what you use, and you can cap it with a monthly spending limit; see what it costs.

Are my API keys sent to Noerion Labs?

Bring your own keys: Your provider keys stay on your PC and are never sent to Noerion Labs.

Can Noerion Labs read my prompts?

Bring your own keys: Valet talks directly to the model provider you configured. No request passes through Valet's servers.

See how your data is handled.

What data does the application transmit?

The application sends no usage or telemetry data.

The website receives only what is listed in what the website receives - including feedback, if you choose to send it from the application.

What changes when the managed plan arrives?

Pro, planned: On the managed plan, requests will pass through Valet's servers to the model you chose, and will not be stored there.

On that plan you won't need keys of your own. The plan you use today is unaffected. See when the managed plan arrives and the plans.

Which actions require my approval?

Every action has a risk level. High-risk actions wait for your approval, and no model can approve or skip them, by design. Sending, posting, paying, sharing and deleting are examples of high-risk actions, and each outgoing message is approved on its own. See how approvals work.

Does it work offline?

Valet depends on your chosen model provider and, for web search, a web-search provider - both need a network connection. A local model, such as one run through Ollama, can reduce that dependence.

Which Windows versions are supported?

Windows 10 or later.

How do I uninstall?

Uninstall Valet the same way as any other Windows application, from Settings or Control Panel.

Troubleshooting

The installer fails

Confirm you're running it with administrative rights and that your Windows version is supported. If it still fails, send us feedback with what happened.

The background service doesn't start

Try repairing the installation from the installer. If the service still won't start, send us feedback.

Model provider authentication fails

Check that the key in Settings > AI Models is complete (copied in full, with its prefix), that it hasn't been revoked, and that your provider account has a payment method or credit. If you set a spending limit, check it hasn't been reached.

A plugin fails to load

Confirm the plugin is still signed and that you're running a supported product version - see the plugin's own listing for its minimum version, where one applies.

Still stuck? Send us feedback.