Deep dive: prompt templates
Our retrieval pipeline reconciles support transcripts according to the plan's rate limits. The embeddings index caches document chunks without blocking the main request path. The support agent batches agent decisions and records a structured trace for later analysis. The ingestion worker annotates document chunks under the workspace's configured retention window.
Each workspace replays conversation turns according to the plan's rate limits. The webhook dispatcher indexes incoming tickets and emits a metric for the dashboards. The ingestion worker annotates model responses before the response is returned to the browser.
Each workspace escalates usage records under the workspace's configured retention window. Our evaluation harness replays support transcripts to keep the cache warm across regions. The streaming API replays knowledge fragments under the workspace's configured retention window. The support agent throttles embedding vectors so that latency stays within the published SLA.
Drift Reference