Show HN: Tessary – Find the AI agent failures your sampled evals miss

Category: observability

Tags: ai-observability, ai-agents, opentelemetry, root-cause-analysis, llm-monitoring

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 7, Utility: 7)

Tessary is an open-source reliability platform for AI agents that monitors every trace, detects failures using cheap classifiers, groups findings into cases, and performs root-cause analysis with trace and repository evidence. It fills a real gap in production AI observability by moving beyond sampled evals to exhaustive trace monitoring with automated triage.

Target audience: AI engineers, platform teams, and devops running LLM agents in production

Repository: https://github.com/tessaryai/tessary · Java · Apache-2.0 · 20 stars

View on Hacker News