Create Experiment Campaign | Sarvam API Docs

Create a campaign that runs an experiment. This uses the same Create campaign endpoint. Send experiment_id on app_config and leave out app_id and app_version. Dial rate, phone numbers, retries, schedule, and the webhook are unchanged — see Create campaign for the full schema.

POST https://apps.sarvam.ai/api/scheduling/v1/orgs/{org_id}/workspaces/{workspace_id}/campaigns

Create the experiment first with Create an experiment.

Request

{
  "name": "Q1 Payment Reminders — greeting experiment",
  "description": "Split the greeting across two committed versions",
  "app_config": {
    "app_type": "agent",
    "experiment_id": "greeting-te-9f1c2a7b-4d3e",
    "attempts_per_second": 2,
    "connection_configs": [
      {
        "connection_id": "conn-xyz",
        "phone_numbers": ["+918041234567"],
        "weight": 1
      }
    ],
    "retry_config": {
      "max_retries": 3,
      "retry_interval_minutes": 30,
      "retry_on": {
        "busy": { "enabled": true },
        "no_answer": { "enabled": true },
        "short_duration": { "enabled": true, "threshold_seconds": 30 }
      }
    },
    "phone_rotation": { "agent": true },
    "webhook_config": {
      "url": "https://your-server.com/webhook",
      "metadata": { "team": "collections" }
    }
  },
  "start_timestamp": "2026-04-01T09:00:00Z",
  "end_timestamp": "2026-04-15T18:00:00Z",
  "allowed_schedule": {
    "timezone": "Asia/Kolkata",
    "allowed_start_time": "09:00",
    "allowed_end_time": "18:00",
    "allowed_days": ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday"]
  }
}

Do not send app_id or app_version with experiment_id.

Response

Field	Value
`app_id`	The agent the experiment is testing.
`app_version`	`null`. The version is chosen per attempt.
`experiment_id`	The experiment you sent.

After create

The campaign stays on that experiment. A later update can change the dial plan when you resend the same experiment_id. It cannot switch the campaign to a single version, or to a different experiment.

For the traffic split and variant rules, see Experiments.