Comparison
Data observabilityvsObservability
Data observability
you get told that a table's volume, freshness and schema look unusual without anyone having written a test for that table.
Automatically monitoring datasets for anomalies in freshness, volume, schema and distribution, across everything rather than where someone thought to look. Its complement is testing, which encodes specific known expectations; observability finds the problems nobody anticipated on the tables nobody owns. It generates far more noise than tests and is worth it mainly at a table count where hand-written coverage has stopped being plausible.
Full entry →Observability
the incident came in and you had every prompt, output, model version, latency and cost to look at instead of guesses.
Logging, tracing and metrics for LLM features: every call's prompt, output, tokens, latency, cost, model version and outcome. Without it, production incidents reduce to guessing. It is the same discipline you apply to services, with prompts and completions as the payload.
Full entry →