Oodle.ai: Agent Observability at $10 per Million Traces
kirankgollu · hn · 2026-07-14
This is a Show HN post for Oodle.ai, an LLM/Agent observability platform priced at $10 per million agent traces.
The authors detailed their approach:
- They built a custom columnar storage engine for logs, metrics, and traces.
- Agent traces are stored entirely in S3 using a proprietary Parquet-like format, queried via AWS Lambda.
- Because agents are highly non-deterministic, they emphasize zero sampling to avoid losing critical traces.
For analytics, they prioritize deterministic processing before running LLM evals:
- Detecting production signals like tool failures, retries, loops, abnormal token usage, latency fallbacks, schema violations, and sentiment.
- Targeting sub-second p99 query latency while maintaining healthy margins.
- They note their costs are 6x lower than when they used Langfuse.
The project is in early stages, and the authors are welcoming feedback.
More from coding & agent
- A better path to agent autonomy is running waves, finding friction, and iterating — JnBrymn · 2026-07-22
- Coding agents are heading toward an AI-writes, AI-reviews, human-approves workflow — aftahi_ai · 2026-07-22
- oMLX 0.5.2 adds Mac menu-bar stats, low-bit decode kernels, and faster downloads — awnihannun · 2026-07-22
- GitHub review bot hits its PR limit and forces a 39-minute cooldown — DanielLockyer · 2026-07-22
- Max reasoning effort appears to be mobile-only in Codex Remote, not desktop — GabGarrett · 2026-07-22
- A Reddit demo argues online stores should expose carts and pricing through MCP — gelembjuk · 2026-07-22