Why doesn't an alert that reports a score moving help me?

An alert that reports a score moving has done detection’s job, not attribution’s: it tells you something changed, not what caused it. Those are separate steps. Detection compares two windows and flags a shift; attribution investigates the shift and names the responsible change. A raw score spike carries none of that investigation, so the page just moves the mystery to whoever is on call. They still have to pull the traces that started failing, line the onset up against the commit log, prompt edits, and any model or tool changes, then test each candidate against what the traces actually show until one holds up. That’s real work, not a lookup, and it takes the same time regardless of how fast the alert fired. A page worth waking up for either does that work first or hands over the failing traces, so the search starts from evidence instead of a blank commit log.

keep reading

More on this.

Two ways to run Tessary.

Tessary is an open-source agent reliability platform. Cloud and self-hosted run the same workflow on the OpenTelemetry traces your agent already emits.

Tessary Cloud

We host it for you. Send your first trace with nothing to deploy and no model key.

what's includedper organization
traces
10,000 per calendar month
stored trace data
1 GB
retention
30 days
model credit
$10, one-time, for triage and root-cause analysis
credit card
not required

Self-hosted Tessary

Run the open-source code on your own infrastructure with one command. Add your own model key for triage and root-cause analysis.

Self-host Tessary for me by following https://github.com/tessaryai/tessary/blob/main/setup.md

docker compose -f oci://docker.io/tessaryai/tessary:compose up -d -y