Engage does not resell AI credits and does not proxy your content through any server of ours. You connect a key from a provider you already trust, and Engage calls that provider directly from your own site.
The three providers:
- Google Gemini, recommended, because Google AI Studio keys have a real free tier and need no card. The default model is gemini-flash-latest, Google's always-current Flash model.
- OpenAI, paid per use. The default model is GPT-5 mini.
- Anthropic Claude, paid per use. The default model is Claude Sonnet 5.5.
The first model in each list is the best value for everyday use. A model you already chose stays exactly as you picked it.
Connecting a key:
- Open the PRO tab. If your licence includes PRO, it is a control panel rather than an upsell.
- Choose your primary provider from the dropdown.
- Click the Get your key link beside the key field. It goes straight to the right page, and there is a 60 second explainer underneath.
- Paste the key and click Save AI Key.
- Click Test Connection. Engage makes one cheap round trip and tells you plainly whether the key works.
Each provider has a Model dropdown with the price per million tokens shown beside every model, and a fast, low cost default already selected. Leave it alone unless you specifically want a different model.
A fallback provider:
Under Fallback provider (optional) you can choose a second provider. If the primary hits a rate limit, an outage or an error, Engage tries the fallback automatically. The fallback needs its own key, which you paste into its own panel.
When a provider retires a model:
If a model you picked has been retired by its provider, Engage automatically uses its current replacement for the request, so the AI features keep working with nothing to change. Your saved choice is left as it is.
Your key is masked as soon as it is stored. It is never sent back to your browser in full, and leaving the masked value untouched means keep the stored key.
What it costs:
With a free Gemini key, the AI features can cost you nothing to run. With OpenAI or Claude you pay that provider directly at their usual per use rates. Engage adds nothing on top either way.
Monthly soft budget:
The Monthly soft budget (USD) box begins at 20. When this month's estimated spend crosses it, the AI Usage panel shows a warning. Nothing is blocked: the hard limit is the one you set with your provider. Set it to 0 to switch the warning off.
The AI Usage panel:
Below your keys, AI Usage shows total jobs, jobs in the last 30 days, this month's estimated spend and errors in the last 30 days, then the last 10 jobs with their type, status, model, estimated cost, how long they took and when. Costs are estimates worked out from your provider's published per-token prices, and your provider's own dashboard is the real bill.
If a request fails you get the provider's own message back in plain English, including the two most common causes: a rejected key, which means copy it again, and a rate limit, which means wait a moment and retry.