Optimize AI costs with Agent Observability

Datadog Agent Observability helps organizations understand and manage AI agent spending by connecting estimated model costs to the applications, traces, workflow steps, and technical behaviors that generate them rather than relying solely on provider bills organized by API key or model. Its Cost page breaks spending down by application, provider, model, token type, environment, and custom tags, while showing metrics such as cost per request, cache hit ratio, and cache reuse ratio to distinguish increased traffic from inefficient processing. Engineers can investigate costly requests through ranked LLM calls, traces, spans, and execution graphs that reveal repeated calls, agent loops, redundant tool activity, retrieval steps, and expensive subtasks, with optional correlation to Datadog APM for broader performance analysis. Cost Savings Opportunities identifies potentially avoidable spending, particularly from ineffective prompt caching, repeated conversation histories, unstable prompt formatting, and changing tool definitions, and provides estimated savings, supporting evidence, realistic targets, and remediation guidance. Teams can validate changes with request-level observations and experiments that compare quality, evaluations, token consumption, and costs, aiming to improve efficiency without degrading outputs; the service is free for customers submitting up to 40,000 LLM spans monthly.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.