Skip to main content

🌐 API calls

1. OpenAI Completion​

  • Action Trigger: When the user presses the Start Record button, the app records their speech and converts it into text.

  • API Call: Once the text is generated, the app sends the speech text to OpenAI's endpoint for a reply.

    prompt

2. Stripe - Checkout Session​

  • Action Trigger: When the user clicks the "Subscribe" button, two actions occur:

    • 1.A Stripe customer is created using the logged-in user’s email.(custom action)
    • 2.A Stripe Checkout session is generated using the created customer ID.

  • API Call: The app sends the text from the previous response to OpenAI's TTS API.

    • Endpoint: https://api.stripe.com/v1/checkout/sessions
    • Method: POST
    • Body:
     {
       "customer": "<CUSTOMER_ID>",
       "payment_method_types": ["card"],
       "line_items": [
         {
           "price_data": {
             "currency": "usd",
             "product_data": {
               "name": "Subscription Plan"
             },
             "unit_amount": 1000
           },
           "quantity": 1
         }
       ],
       "mode": "subscription",
       "success_url": "<YOUR_SUCCESS_URL>",
       "cancel_url": "<YOUR_CANCEL_URL>"
     }
    • Response: The API returns a session informations, which is used to redirect the user to the Stripe-hosted checkout page.
Response customization...

You can customize bot responses in two different ways, depending on your needs:

  • Basic Customization: you can customize the entry prompt by adding any additional information you need, like in the example below: prompt

This way, your bot will always:

  • generate a response that could be read in about 13 seconds.
  • end its response with a small question to keep the conversation going.
  • introduce himself as "RapidAsk bot".
  • Advanced Customization: OpenAI lets you fine-tune models with your own data to tailor responses to your specific needs. For example, if you’re building a medical bot, you can train the model using your medical FAQs or answers to improve its accuracy. After uploading your dataset (ex: text, pdf, media, links...) and training the model, you can access it via an API endpoint.

more about fine-tuning docs.