RCA is the explain step at the end of Tessary's pipeline. By the time it runs, the cheap classifiers have swept every trace and recorded findings, triage has ruled which of those findings are real deviations rather than legitimate changes, and the real ones have become a case. RCA runs on that case. Its job is attribution: saying why the deviation happened.
Each run is an agentic session in a fresh E2B microVM with the connected repo. It investigates the failing cohort with three inputs: the trace evidence, a dossier of what's known about the case, and the code itself.
The output is a set of grounded hypotheses for the cause, each with the evidence behind it, plus the checks the investigation ran and ruled out, so engineering doesn't repeat them. The report cites the files involved.
After attribution, RCA raises a PR with the fix.
8 questions
Answered, plainly.
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.
- 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