Tag
I implemented a minimal generate-evaluate-feedback-regenerate loop in a verification script. This post organizes the stop conditions and evaluation units that actually matter when stabilizing AI output.
A hands-on report instrumenting Sakana Fugu's OpenAI-compatible API with Langfuse, measuring how latency, token consumption, and TTFT change across Level 1–3 tasks.
I subscribed to Sakana Fugu to understand its nature as an OpenAI-compatible API and to plan how to observe its black-box cooperative reasoning from the outside.
An experiment log where I redact Flue 1.0 Beta observe events before sending them to Langfuse, tracking the issue triage workflow's runId, model, and results.
A practical experiment that aggregates Langfuse traces in a Cloudflare Worker and sends only token, cost, and latency anomalies to Slack.
Manually checking traces every day is unsustainable. A personal GenAIOps plan to aggregate Langfuse failures, token spikes, and low-score outputs into a single Slack morning briefing.
I explain why output quality stays unstable even with careful prompt design, and how I switched to a generate-evaluate-feedback-regenerate loop. Includes the smallest manual steps to start today.
Using Cloudflare AI Gateway as an OpenAI-compatible endpoint, I walk through its logging, payload controls, metadata, cost estimation, and OTel integration to see how far it can serve as an entry point for LLM observability.
The minimum steps to start using Cloudflare AI Gateway from Workers AI, and what to understand about its role, settings, and checkpoints before moving to the OpenAI-compatible API.
As tools like Claude Code and Codex accelerate code generation, the scope of what engineers must design expands from the code itself to problems, constraints, operations, and validation.