Observability
Observabilitas
Being able to see inside a running system well enough to explain what it did.
Observability is the ability to understand a system from its outputs, without having to guess at its internals. In a classical service that means logs, metrics, and traces. In an AI system it also means recording the prompt, the model id, the retrieved passages, the tool calls, and the token counts, because that is the only way to explain a specific answer later.
The AI-specific part is traceability. When a user asks why the system said that, the answer lives in the record of that particular call, and it is only there if you saved it. Logging the request without the response, or the response without the model version, is not enough to reconstruct anything.
Keep the records long enough to be useful and short enough to be affordable, and make sure they are structured. A log you cannot query is a file you will not read.
Related terms
Monitoring
Watching a running system so you learn about drift before your users complain.
Deployment
Getting a working system in front of real users, reliably, and keeping it there.
Regression
A change that made something worse. The reason to keep tests at all.
Evaluation
Measuring whether the system does the job, on a fixed set of cases, repeatedly.
Trace
The recorded path of one request through the whole system.
Want this working inside your team?
It starts with one conversation about your workflow.