What does the AI cost?
AnswerKite uses your own OpenAI API key, so OpenAI bills your account directly – there is no markup and no message limit from us. With the default model, gpt-4o-mini, one answer typically costs a fraction of a cent. Building the knowledge base of a typical small site costs less than a cent, and later updates only process what changed.
You can choose another Chat model under Settings ▸ Connection (for example gpt-4o or gpt-4.1 for higher quality at a higher price). Set a monthly budget limit in your OpenAI account if you want a hard cap.
What is stored
- The visitor’s name, email and the conversation are stored in your own WordPress database (AnswerKite ▸ Conversations).
- Conversations are deleted automatically after the number of days under Settings ▸ Privacy & advanced ▸ Delete conversations after (0 keeps them).
- IP addresses are only stored as a salted hash, for rate limiting.
- Your OpenAI key stays on your server and is never sent to visitors’ browsers.
What is sent to OpenAI
To write an answer, the question and the most relevant sections of your site are sent to OpenAI. Only published, non-password-protected content is used. Mention this in your privacy policy, for example:
Our website chat is powered by AnswerKite and OpenAI. When you use the chat, your name, email address and messages are stored on our website for [number] days, and your messages are sent to OpenAI to generate answers.
The Privacy note setting shows a short notice with a link to your privacy policy under the chat’s name and email form.
Other AI services
Under Settings ▸ Privacy & advanced you can change the API base URL to use an OpenAI-compatible service or proxy. Changing the Embedding model requires a knowledge base rebuild.