Whet pixel markWhet

Acceptable use and AI features

What Whet may be used for, and how its AI models run

The few things Whet may not be used for, what its AI features may not be used to generate, and a plain account of the AI features: which models run where, what each one is sent, and whose contract covers it. This page is part of the terms and applies to the trial and to a licensed copy alike.

1. What Whet may not be used for

You may not use Whet, or let anyone use your copy:

Whet may be used for safety-related engineering. Section 9 of the terms still applies: the results are yours to check, and Whet is not certified for any safety-critical purpose.

2. What the AI features may not be used to generate

The assistant answers questions and writes text, code, scripts and sketches on request, and Whet also generates renders and video. Whichever model does the work, local or one you attach, you may not use these features to generate, or try to generate:

These rules apply in addition to the usage policy of the model or provider you use, and they apply even where that model would produce such content. The seller does not see what you generate, so keeping to this section is your responsibility.

3. The AI features, in one paragraph

Whet's assistant is an AI model, not a person. By default it runs on your own machine, through a model runtime you install, and nothing you ask it leaves the machine. You can also point it at a second machine of yours, or attach cloud models with your own API keys. The seller runs no AI service, receives none of your prompts or answers, and trains no model on anything of yours, because none of it is ever sent to the seller.

4. Each way a model can run

Local models, on your machine
Whet asks the runtime you installed (for example Ollama) to run a model and talks to it on your machine's own loopback address. The model files are downloaded by that runtime from its own registry, under its own terms, and each model is under its own licence; the default one is Qwen3, under Apache-2.0. Whet does not ship model weights in its installer.
A second machine of yours
If you name another machine for models, Whet sends the questions for the model to it over ssh, with the ssh program and keys already on your computer. That machine, its security and your right to use it are your responsibility.
Cloud models you attach
If you attach a provider such as Anthropic, OpenAI, Google Gemini, xAI, DeepSeek, Groq, Mistral, Together or OpenRouter with your own API key, Whet sends that provider the questions you route to it. That is a contract between you and the provider: its terms, its usage policy, its privacy policy and its charges apply, and the seller is not a party to it. Whet keeps your keys on your machine and sends each one only to the provider it belongs to. A provider you have not attached is never contacted.

5. What you are responsible for when you attach a provider

6. AI answers can be wrong

A model can be wrong, including confidently. Whet checks the figures a model states against the solver's own records where a record exists, and says when it cannot check one. That reduces the risk; it does not remove it. Section 10 of the terms applies: what you decide with a model's help, and what you do with its output, remain yours. Output a model writes for you may resemble existing material; check it before you publish or rely on it.

7. If this policy is broken

A serious or repeated breach is a breach of the terms, and section 18 of the terms says what can follow. To report misuse of Whet, write to [email protected].

Change log

This version was published on 2026-10-31.