Yeja Studios
Sign In Join the Waitlist

AI

AI & Automations

Connect your own AI provider keys, try models in Test Chat, and send signed webhooks to your server when you make a sale.

Where To Find It

Open the account menu and choose AI & Automations, or open the AI & Automations card in the Account Hub. The pane has three views: AI Providers, Webhooks and Test Chat.

Connect Your Own AI Provider

Yeja uses your own provider accounts for its AI features, so you bring your own key. Each provider bills you directly; Yeja never pays for or resells your usage. Any signed-in account can connect providers and use Test Chat.

  • Choose Connect Provider, pick a Provider, give it a Connection Name and paste your API Key.
  • Supported providers: OpenAI, Anthropic, OpenRouter, Google Gemini, DeepSeek, Meta, Mistral, xAI and Groq.
  • Choose Verify & Save. Yeja checks the key by asking the provider for its model list. It sends no prompt and generates nothing.
  • Your key is encrypted on Yeja servers and never shown again. The list shows only its last four characters.

Manage a Connection

  • Choose Edit on a connection to rename it.
  • Choose Load Models, then pick a Default Model for Yeja AI features as they arrive.
  • Choose Replace Key to paste a new key. Yeja verifies it with the provider before it replaces the old one.
  • Choose Remove Connection to delete the encrypted key. Revoke the key in your provider account too if you no longer use it.

Try Models in Test Chat

Test Chat sends prompts through your own key, and every reply shows its tokens, estimated cost and time. Use Compare to send one prompt to up to three models at once. Set Max Reply and an optional System Prompt, then choose Save Settings to open Test Chat the same way next time. The Report panel totals this chat, estimates use cases on your chosen models and shows the Last 30 Days. Yeja doesn't save the text of your prompts or replies; it records only tokens, cost and time for the report. Estimates use published prices, and your provider’s invoice is final.

  • Save Settings keeps up to three setups: the models, Single or Compare, Max Reply and System Prompt. They are listed under Saved Settings in the Report, where you can Load or delete one. Only settings are saved, never the conversation.
  • Reset clears the chat and goes back to the saved settings you loaded, or to the defaults.
  • When you pick a model, Common Use Cases suggests what it is usually chosen for, from its family, price and features, with Test With buttons. It is a starting point; your own tests are the real answer. Not a Chat Model means Test Chat can’t send text to it.

Price Yeja Use Cases

The Use Cases tab of the Report shows what four Yeja AI tasks would cost on the models you chose, per run and per 100 runs:

  • Artwork Description & Alt Text: a short public description and screen-reader alt text for one artwork.
  • Listing Title & Tags: a sale title and search tags for a listing.
  • Collector Match Note: why one artwork fits one collector’s taste, in two sentences.
  • Translate Your Bio: your artist bio in another language, keeping your voice.

Choose Try It to put a sample prompt in the chat, then send it with your own key. Costs start from Typical sizes and switch to Measured once you have tried a use case. Estimated Value isn’t available yet and isn’t one of these use cases.

Send Webhooks to Your Server

Webhooks need an active Pro subscription. Without one, the Webhooks view shows See Plans. See Pricing.

  • Choose New Webhook, enter an Endpoint URL that starts with https://, pick your Events and choose Create Webhook.
  • New Sale is sent today. Order Refunded and Subscription Updated can be selected but are marked Not Sent Yet.
  • Copy the Signing Secret right away, because it is shown only once. Every delivery carries an X-Yeja-Signature header that your server can check with it.
  • Open a webhook with Edit to change its URL or events, switch it off with Enabled, choose Send Test, or choose Rotate Secret. Rotating stops the old secret at once.
  • Recent Deliveries lists the latest deliveries with their status: Delivered, Failed or Pending.
  • You can have up to five active webhooks.

Common Questions

Do I need Pro to connect an AI provider?

No. Any signed-in account can connect providers and use Test Chat. Only webhooks need an active Pro subscription. See Pricing.

Why was my key rejected?

Check that you picked the matching provider, copied the whole key, and that the key is active and allowed to list models. If the provider returns no models, check API access and billing in your provider account. You can check up to five keys an hour.

Who pays for Test Chat messages?

Your provider bills your own account directly. Test Chat shows an estimate from published prices; your provider’s invoice is final. You can send up to 30 test messages an hour.

I lost my signing secret. What now?

A secret is shown only once. Open the webhook, choose Rotate Secret, copy the new secret and update your server. The old secret stops working at once.

How many connections and webhooks can I have?

Up to 12 provider connections and up to five active webhooks. To add a sixth webhook, switch one off first.

How many Test Chat setups can I save?

Up to three. To save another, delete one under Saved Settings in the Report first.

What does Measured mean on a use case?

You have tried that use case in Test Chat, so its cost uses your own average token counts instead of typical sizes.

Can my AI provider estimate an artwork’s value?

No. Estimated Value isn’t available yet, and it isn’t offered in Test Chat.