SETUP
1. Create a free account at retellai.com and confirm your email.

2. In the Retell dashboard, open Agents and create a voice agent.
  Copy its Agent ID (starts with "agent_").

3. In the Retell dashboard, go to Settings > API Keys and copy your key
  (starts with "key_").

4. In your Bubble app, open Settings > Plugins > Retell AI Speech Assistant.
  In the "Authorization" field enter your key in this exact format,
  including the word Bearer and one space:

      Bearer key_xxxxxxxxxxxxxxxxxxxx

5. Drag the "Retell Voice Agent" element onto a page.

6. Build the start workflow:
  a) Add a Button, then Start/Edit workflow.
  b) Step 1 — Plugins > "Create Web Call". Set agent_id to your Agent ID
     from step 2. This returns an access_token.
  c) Step 2 — Element Actions > "Retell Voice Agent - Start call".
     Set Access Token to: Result of Step 1's access_token.

7. Build the stop workflow: a second Button running
  Element Actions > "Retell Voice Agent - Stop call".

8. Preview the page, allow microphone access when the browser asks,
  and click Start. You are now talking to your Retell agent.

READING THE CONVERSATION
The element exposes these states while a call runs:
 is_calling      - yes while a call is active
 agent_talking   - yes while the agent is speaking
 transcript      - the live conversation transcript
 last_error      - the most recent error message

And these events: call started, call ended, agent talking,
transcript updated, an error occurs.

NOTES
- You need your own Retell account. Retell's usage pricing applies.
- The API key is stored server-side and is never exposed to the browser.
- The plugin stores no user data.

DEMO
Live demo: https://emrejsx.bubbleapps.io/version-test
Demo editor: https://bubble.io/page?id=emrejsx&tab=Design&name=index