Tessary

Stop your agents from failing silently in production

Visit website

Open-source tool that monitors AI agent traces, groups failures into cases, and finds the root cause in your codebase.

Tessary homepage hero

Tessary homepage hero

1 / 2

Overview

What is Tessary?

Stop your agents from failing silently in production

Tessary is an open-source agent reliability platform for teams running AI agents in production. It watches every production trace (no sampling), runs cheap classifiers to catch failures, behavior drift, and user frustration, groups validated issues into cases for triage, and performs agentic root-cause analysis against your agent codebase — prompts, tools, models, and agent code — reporting causes with evidence.

Tessary is free to start: self-host the Apache 2.0 open-source stack with Docker, or use Tessary Cloud with a free tier of 10,000 traces/month, 30-day retention, and a $10 one-time triage/RCA credit — no credit card required. Paid plans are custom quotes with custom classifiers, hosted instances, higher limits, longer retention, and Slack support.

Get started by sending traces to the OTLP endpoint with OpenTelemetry and tagging model-call spans with tessary.call_site.id.

Platforms and languages

Tessary Availability

Platforms

WebApi

Languages

English

Capabilities

Tessary Key Features

Full observability

Monitors every production trace as it lands — no sampling, no extra LLM calls, just your existing OpenTelemetry spans.

Cheap classifiers

Runs narrow, low-cost classifiers on every trace to detect failures, behavior drift, cost spikes, and user frustration.

Cases for triage

Validated issues are grouped into a case so your team can triage, investigate, and resolve them in one place.

Root-cause analysis

Runs agentic RCA against your agent codebase — prompts, tools, models, and agent code — and reports the cause with evidence.

Best for

Who uses Tessary?

Engineering teams running AI agents in production

Monitor every production trace, catch silent failures and drift, and fix them with root-cause analysis

Plans and access

Tessary Pricing

Freemium

Free tier: 10,000 traces/month, 30-day retention, no card required. Paid plans are custom quotes.

Free trial available

Common questions

Tessary FAQs

What is a "Case" in Tessary?

When Tessary's classifiers flag something, it becomes a finding. Findings you confirm move into triage, then get grouped into a case — one place to investigate and resolve the underlying issue.

How do I connect my agent?

Send your production traces to Tessary's OTLP endpoint using OpenTelemetry, and tag your model-call spans with tessary.call_site.id so failures can be attributed to code.

Is there a free plan?

Yes — self-host the open-source stack, or use Tessary Cloud's free tier (10,000 traces/month, 30-day retention, $10 one-time triage/RCA credit, no credit card).

Can it alert my team?

Yes — alerting routes through webhook, PagerDuty, Sentry, and Linear, with quiet hours so on-call isn't spammed.

What does it check out of the box?

Five default classifiers cover duration and cost drift, tool errors, secret leaks, and malformed output. Paid plans add custom classifiers.

Reviews

See what the community thinks and share your experience.

—

Based on 0 ratings

Rating distribution

0
0
0
0
0

Leave a review

Sign in to rate

Community reviews

No reviews

No written reviews yet.

Explore by category

AI Monitoring Tools

View all AI Monitoring Tools websites

Explore by category

AI Development Infrastructure

View all AI Development Infrastructure websites