Usage, credit and model keys

Hadir AI runs on AI models, either on Hadir-managed keys (the default) or on your own model provider keys if your plan includes that. Everything here lives in the inbox under Settings, then Hadir AI settings, a page for administrators only.

How replies are counted

Your plan includes a number of AI replies each month. What Hadir AI writes on Hadir-managed keys counts towards them: answers it gives your customers, drafts and suggestions for your team through HadirPilot and the reply tools, voice notes it transcribes, and FAQs it writes from your documents.

You can follow usage in two places:

• In Hadir AI settings, under Model keys (1), a line such as "This month: … of … AI replies on Hadir-managed keys."
• In your account area, the Hadir AI this month card on the Overview page. See Your account area.

When the month's replies run out, Hadir AI stops rather than running up a cost: the assistant hands conversations to your team, HadirPilot and the Playground tell you so, and voice notes are not transcribed until the new month, unless your plan includes prepaid credit.

AI credit (owner only)

If your plan includes pay-as-you-go replies, replies past the ones included in your plan are paid from a prepaid balance. The balance is also the cap: when it runs out, Hadir AI hands conversations to your team, and nothing is ever charged beyond what you added.

Only the workspace owner can see and top up the AI credit, so other administrators do not see the credit card on the page.

1. In Hadir AI settings, select Open wallet on the AI credit card.
2. You see the Balance (1), the price per reply past the included ones, and how many included replies this period has used.
3. Under Add credit, pick an amount (2), and a payment method under Pay with if it appears.
4. Select Add credit (3). The checkout opens in a new tab; the credit is added the moment the payment is confirmed, and the page updates when you come back.

Recent movements lists every Top-up, every Reply charged and any Adjustment by Hadir. Invoices and receipts are in your account area under Plan and billing.

Run Hadir AI on your own keys

If you have an account with a model provider, such as OpenAI, Anthropic or Gemini, you can run Hadir AI on it. The provider bills you for usage directly, and this usage does not count towards your plan's replies.

1. In Hadir AI settings, under Model keys, switch on Use my own model keys (2). If the switch is disabled you will read: "Bringing your own model keys is available on a higher plan. Change your plan from your account dashboard."
2. Choose the Provider and the Chat model from the list, or choose "Other — type a model name" and type it.
3. Paste the API key. It is stored encrypted and never shown again; only its last four characters are.
4. Base URL is optional, unless the provider requires it.
5. Under Embeddings (knowledge search), leave the Embedding provider as "Same as chat provider / Hadir-managed", or pick another provider if yours has no embedding models.
6. Select Test and save. Hadir AI makes one tiny test request with the key, and a key that does not answer is not saved. On success you see "Key verified and saved."

Under Model per feature you can also run a feature on a different model from the same provider; an empty field means your chat model.

To go back to Hadir-managed keys, switch the setting off; your keys are kept. To delete them, select Remove and use Hadir-managed keys.

Features

At the bottom of the page, under Features, you switch Label Suggestion and Audio Transcription on or off. The second is covered in Voice-note transcription.

Note: if you read "Your plan includes the Hadir AI pages but no model keys", your plan shows the Hadir AI pages but includes no keys to answer with, so Hadir AI will not reply until you change your plan.