Exa Observability

Overview

Monitor Exa AI search API performance, result quality, and cost efficiency. Key metrics include search latency (Exa neural search typically takes 500-2000ms), result relevance (measured by click-through or downstream usage), search volume by type (neural vs keyword vs auto), per-search cost tracking, and cache hit rates for repeated queries.

Prerequisites

• Exa API integration in production
• Metrics backend (Prometheus, Datadog, or equivalent)
• Request logging infrastructure

Instructions

Step 1: Instrument the Exa Client

import Exa from 'exa-js';