How to connect FreeScout to Google Gemini
Works with Google Gemini
In ReplyRabbit, Gemini is configured through the Google AI Studio provider. Open a mailbox's ReplyRabbit settings, choose Google AI Studio, paste an API key from AI Studio, pick a Gemini model, and save. Gemini then writes that mailbox's drafts. Google AI Studio is a paid-plan provider, so the mailbox needs ReplyRabbit Pro or Team. If you're looking for "Gemini" in the provider list and can't find it, that's the reason: look for Google AI Studio.
Where Gemini fits
Google's Gemini line includes fast, low-cost models alongside larger ones, which makes it a common choice for high-volume mailboxes and for background tasks such as mailbox previews. Whether a Gemini model matches your tone is something to test on your own tickets; ReplyRabbit lets you run it on one mailbox while another stays on your current provider.
Requirements
- ReplyRabbit Pro or Team for the mailbox.
- A Google AI Studio account and an API key from it. Google bills usage to that account.
- The FreeScout server must be able to reach Google's API endpoint. If outbound traffic is restricted, allow it before testing.
Connect Google AI Studio
Open the mailbox settings
In FreeScout, open the mailbox, choose Settings, then ReplyRabbit.Choose Google AI Studio
In the AI Provider card, set Provider to Google AI Studio.Paste the API key
Paste the AI Studio key into API Key. It's stored as a secret and shown masked afterwards.Choose a Gemini model
Pick the latest option or a listed Gemini preset. Use Custom only with a model ID you've confirmed your account can use.Save and test
Save, open a customer email in the mailbox, and generate a draft. Check the tone against your response style before rolling it out to a busy inbox.
Using Gemini for the cheap jobs only
You don't have to move your drafts to Gemini to benefit from it. In the AI Provider section, the Mailbox previews and Attachment vision subsections (both Team features) can point at Google AI Studio while Drafts stays on whatever writes your best replies. Previews run far more often than drafts, so a low-cost model there saves the most. The mechanics are in Per-task AI models.
Caching and budgets
ReplyRabbit doesn't force a cache setting through the Gemini route. When Google returns cached-token metadata on a response, ReplyRabbit records it with the usage data. The mailbox's Monthly Token Budget counts total input and output tokens either way.
When drafts look wrong
| Symptom | What to check |
|---|---|
| Google AI Studio is disabled in the list | The mailbox is on the free plan; upgrade to Pro or Team. |
| Key rejected | Re-copy it from AI Studio and confirm the project has API access enabled. |
| Model error | Try the latest option; a custom ID may not be available to your account. |
| Replies are terse or generic | Fill in company context, response style and a few examples, then regenerate. |
| Provider unreachable | Confirm the server can make outbound HTTPS requests to Google's API. |
General fixes live in Troubleshooting ReplyRabbit. To weigh Gemini against the other options, read Choosing an AI provider.
Official references
Provider names, model lists, pricing, privacy terms, and legal guidance change. Check the current source before setting a production policy.