Vapi doesn’t natively handle fine-grained concurrency/load balancing per assistant, so most teams end up managing batching, queueing, and throttling externally through the API. A common setup is using a CRM + queue system together (like HubSpot, GoHighLevel, Airtable, Redis queues, or BullMQ) to control call distribution and retries across assistants. How many concurrent outbound calls are you trying to run per assistant right now?