# What is a call site?

A call site is a place in your code where an agent with a specific function runs: a single independent LLM call, or a whole agent, a prompt plus its tool calls and the loop around them. Either way, it's the unit Tessary binds findings, baselines, and regressions to, because that's the granularity where a fix actually happens. Tessary resolves a call site from one thing only, the `tessary.call_site.id` attribute you set on a span, nothing else. Untagged spans still ingest, they just don't get a call site. Two call sites doing the same kind of work, say, summarizing two different documents, are still tracked separately if they're different places in your code, since the id names the function, not the request. From then on, everything Tessary learns about that function accrues to that one place: its normal latency, its normal cost, its normal behavior.

---

Source: https://tessary.ai/answers/tessary-call-sites/what-is-a-call-site
More on Tessary call sites: https://tessary.ai/answers/tessary-call-sites
From Tessary, agent reliability for AI agents in production: https://tessary.ai
