Quickstart

The r-mad.ai API covers OpenAI-compatible chat completions and audio transcription (with intent/BANT/NEAT-P extraction), both authenticated with the same API key issued to your organization. (Genesys Cloud's Summarization Connector and AudioHook streaming-transcription connector are separate, dashboard-configured integrations — see Authentication and their API references if you're troubleshooting one.)

1. Get an API key

Sign in to the dashboard, open API Keys, and create a key. The full secret is shown once — store it somewhere safe, since only the last four characters are retrievable afterwards.

2. Make a request

return_markdown is required on every request — set it to true if you want the reply formatted as Markdown (headings, lists, code blocks, etc.), or false for plain prose. A request that omits it gets a 400.

curl https://api.r-mad.ai/api/llm/chat/completions \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "user", "content": "Say hello in one sentence."}
    ],
    "return_markdown": false
  }'

import requests

response = requests.post(
    "https://api.r-mad.ai/api/llm/chat/completions",
    headers={"X-API-Key": "YOUR_API_KEY"},
    json={
        "messages": [
            {"role": "user", "content": "Say hello in one sentence."},
        ],
        "return_markdown": False,
    },
)
print(response.json()["choices"][0]["message"]["content"])

const response = await fetch("https://api.r-mad.ai/api/llm/chat/completions", {
  method: "POST",
  headers: {
    "X-API-Key": "YOUR_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    messages: [{role: "user", content: "Say hello in one sentence."}],
    return_markdown: false,
  }),
});
const data = await response.json();
console.log(data.choices[0].message.content);

3. Read the response

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1735689600,
  "choices": [
    {
      "index": 0,
      "message": {"role": "assistant", "content": "Hello! Hope you're having a great day."},
      "finish_reason": "stop"
    }
  ],
  "usage": {"prompt_tokens": 12, "completion_tokens": 11, "total_tokens": 23}
}

That's it — no model field to choose or manage, no separate tool-configuration step. Your org's system prompt (if any) and MCP integration (if any) are both applied automatically.

4. Transcribe audio

Same key, a different endpoint — sent as multipart/form-data rather than JSON, since it takes a file:

curl https://api.r-mad.ai/api/transcribe \
  -H "X-API-Key: YOUR_API_KEY" \
  -F "file=@call.wav" \
  -F "language=en"

import requests

response = requests.post(
    "https://api.r-mad.ai/api/transcribe",
    headers={"X-API-Key": "YOUR_API_KEY"},
    files={"file": open("call.wav", "rb")},
    data={"language": "en"},
)
print(response.json()["text"])

The response includes the transcript, timestamped segments, and a customer-intent classification by default; pass extract_bant=true and/or extract_neat=true to also get a BANT and/or NEAT-P sales-qualification record — they're independent frameworks, not alternatives, so you can request either, both, or neither. Billed separately from chat completions, in audio-minutes rather than tokens — see Rate Limits. Valid language codes are listed at GET /api/languages (unauthenticated — safe to call from anywhere).

Next steps

See Authentication, Rate Limits, and System Prompts for the details, or jump straight to the API Reference for chat completions, transcription, streamed token-by-token replies over Realtime Chat, or live call transcription over AudioHook.