1. What AI does in Hostly
AI does two jobs inside Hostly, and only two.
- It drafts replies. For each review, the AI writes a suggested reply in your voice, so you have a strong starting point instead of a blank box.
- It spots themes. The AI reads across your reviews and surfaces patterns, for example slow service on weekends or a dish that keeps getting praised.
That is the whole of it. AI does not decide what gets published, does not set your prices, and does not make any decision that affects you or a reviewer on its own.
2. Which AI we use
Hostly uses Claude, the AI model made by Anthropic, through Anthropic's Claude API. We chose it because it writes well, follows instructions closely, and comes from a provider with clear commitments about not training on data sent through its API.
3. What data we send it
To draft a reply or find a theme, Hostly sends the following to Anthropic's Claude API:
- Your review text. The content of the public Google reviews of your restaurant, including the reviewer's public display name.
- Your brand voice settings. The tone and preferences you set so the reply sounds like you.
- Your business name. So the reply is specific to your restaurant.
We do not send your card number, which we never have, and we do not send anything the AI does not need to do the job.
4. A human approves every output
Every reply the AI drafts waits for a person to approve it. That person is you.
Hostly holds each draft in your queue. You read it, edit it if you want, and decide whether to use it. When you approve one, you copy it and paste it onto your Google listing yourself. Hostly does not post to Google, and no AI-written text reaches the public without your tap. There is no mode where the AI publishes on its own.
5. We do not train models on your data
Your data is used to serve you, not to train AI.
We do not use your reviews, your replies, or your business information to train, fine-tune, or improve any generalized AI model, and we do not permit Anthropic or any other provider to do so with the data we send them. Data goes to Claude so it can return a draft to you, and that is the extent of it.
6. AI can be wrong
AI makes mistakes. A draft can state something untrue about your restaurant, misread a review, use the wrong tone, or invent a detail. The theme analysis is a helpful summary, not a verified report.
Treat every draft as a suggestion to check, not an answer to trust. Read before you approve. Once you approve a reply, it is yours and you are responsible for it, as our Terms of Service explain.
7. Telling your customers
Whether and how you tell your customers that AI helped draft a reply is your call, and it may depend on the rules of your platform or your local laws. We built Hostly so the final words are always chosen by a person at your restaurant, which is the honest way to use a drafting tool.
8. If we change the AI we use
If we ever add or switch to a different AI provider, we will hold the same standards: a human approves every output, and your data is not used to train models. If that changes anything material about how your data is handled, we will update this policy and our Privacy Policy, and we will email you before it takes effect.
9. How to reach us
Hostly, a product of Vero Dawn
PO Box 5093
Mooresville, NC 28117
United States
hello@hostly-reviews.com
Any question about how we use AI, send it here and a real person will answer.