The GlassFlow blog

A software factory turns a plan into shipped, reviewed software. See how we run ours with open-source Tares and Claude Code agents. Get started in minutes.

We ran Cloudflare's Clef-flash next to Claude Haiku for 6 days on production logs. As a filter, it cut LLM calls ~80%. Test it on your own data with Tares.

Fifteen signals and a five-stage framework for detecting silent AI agent failures while runs are still running

How to debug an AI agent in production: find the run, trace it to the first divergence, classify the failure mode, and ship a fix. A step-by-step guide." slug: "blog/what-is-agent-observability

The seven ways AI agents fail in production: tool misuse, context loss, goal drift, retry loops, frozen agents, cascading errors & silent degradation.

When an AI agent fails in production, it's usually reasoning correctly over stale, partial, or lossy inputs, not hallucinating. Why, and how context fixes it.

The modern data stack was built for human analysts. AI agents read, push, and shape the model themselves. Here's why they need a new kind of data layer.

What it really costs to self-host Langfuse, Phoenix, or Laminar: ClickHouse and infra requirements, engineering hours, and real AWS pricing, single node vs. high availability

The barrier to autonomous agents isn't trust. It's the missing infrastructure. Here are the six pieces the short-session era never built for long-running agents.