Tech news from the best sources
Last week I wrote about our agents fabricating "done" five times in 17 days and the boring external checks that reduced it. This is the embarrassing s…
A signup regression after a Next.js release should not start with panic. It should start with a narrow review. The dangerous version is familiar: a de…
TL;DR: If you only collect metrics, Prometheus Agent mode is lightweight, familiar, and difficult to beat. If you collect metrics, logs, or traces tog…
Legacy frontend-backend deployments fail in layers. A React build can be correct while the proxy serves last week's files. Django can be healthy while…
The Problem with Click-Ops Dashboards Your team has 200 dashboards. You don't know who owns them. Half are broken. The rest show yesterday's reality. …
Where Copilot's telemetry stream lands, decided centrally GitHub added a control on July 8 that lets an enterprise mandate where the Copilot Chat exte…
Multi-step LLM agents fail in a way normal backends don't. Nothing crashes. The pipeline "works", the answer is just bad, slow or three times more exp…
An incident bot caught the CrashLoopBackOff at 3:12 a.m., proposed delete_pod, and the on-call approved it half asleep at 3:14. The new pod went Runni…
Great point @alexshev — downloads ARE a vanity metric. The real signal is: did the agent connect, complete a workflow, and self-diagnose failures? We …
LinkedIn Draft — Insight (2026-07-10) End of week. Here's the thing I kept coming back to: SLOs work when they create conversations, not when they cre…
You Don't Need Datadog (Yet) I see startups spending $5,000/month on Datadog with 8 engineers. That's $625 per engineer per month for monitoring. At t…
Your agent works in testing. Then you deploy it and things break in ways you didn't expect. Here are five failure modes I've seen repeatedly, with Typ…
On March 3, 2026, Helicone announced it was joining Mintlify. If you run Helicone in production, the practical question is not whether the acquisition…
A user reports a hallucinated answer in prod. To fix it you need the full trace of that one request, and how fast you can pull it depends entirely on …
«Отдельные базы больше не нужны», «конец пайплайнов» - каждую неделю кто-то крупный со сцены хоронит то, что ты вчера поставил в прод. ClickHouse пост…
If you've shipped a traditional backend service, you already know the observability checklist: logs, metrics, traces, alerts. LLM-powered apps need al…
Локальная LLM может тормозить и галлюцинировать при полностью «зелёном» мониторинге. Рассказываем, что мониторить помимо инфраструктуры, как устроена …
In modern software engineering, traditional monitoring — simply knowing if a system is up or down — is no longer enough. High-velocity engineering tea…
Six weeks ago, a LangChain agent we'd deployed for a B2B client started failing on roughly 30% of its sessions. No exceptions. No 500s. Nothing in the…
Agent observability gets weirdly polite at the exact moment it should get nosy. It records the model call, stores the prompt, counts tokens, and then …
LinkedIn Draft — Insight (2026-07-03) Something I wish someone had told me five years earlier: Distributed tracing: the gap between having it and usin…
Команда VK Cloud перевела разбор запуска self-hosted (размещаемого на собственных мощностях), read-only ИИ-агента внутри кластера Kubernetes, где всю …
Book: Decoupled PHP — Clean and Hexagonal Architecture for Applications That Outlive the Framework Also by me: Thinking in Go (2-book series) — Comple…
The $50K/Month Monitoring Bill I audited our monitoring stack last quarter. The total cost across all tools: $52,000/month. For a company with 200 eng…
Недавно Claude уверенно пересказал мне большой документ, хотя прочитал только его начало. Я подключил mcp-server-fetch и попросил агента извлечь из до…
Your agent calls a tool. The tool times out at the network layer but actually succeeds on the server. Your harness sees no response, so it retries. No…
$12,000/Month for Logs Nobody Reads Our logging bill was $12,000/month. We were ingesting 2TB/day. When I asked the team what percentage of logs they …
TL;DR — A Kafka + Flink + OTel ingestion pipeline cost us ~$700–800/month at 10 MB/s. We rebuilt it as a single binary where the data, the write-ahead…
For full observability in FastAPI you need 6 pip packages + 60 lines of config + manual glue between logs/spans/metrics. In Fitz it's two decorators a…
На демо AI‑агент может выглядеть надёжным: вызвать инструменты, собрать ответ и отчитаться об успехе. Но в продакшене быстро …