# Insights - [Vector retrieval in production: when a vector database earns its cost](https://larsima.com/hr/insights/vector-retrieval-in-production) — 2026-09-30 — A vector database is an architectural decision, not an automatic upgrade to search. If you can't name exactly what you're missing from plain text search, you probably don't need one yet. - [Typed contracts between backend and client: why shared schemas cut integration bugs](https://larsima.com/hr/insights/typed-contracts-backend-client) — 2026-09-30 — "It worked in Postman" is a sentence nearly every team has said at least once, right before discovering the real problem was somewhere else. A typed contract between backend and client moves an entir… - [Scoring news trustworthiness: designing a system that doesn't claim to be neutral](https://larsima.com/hr/insights/scoring-news-trust-without-claiming-neutrality) — 2026-09-30 — A trust label is an editorial decision encoded in software, not a measurement. Every design choice downstream of that fact — from data model to what you show the reader — depends on accepting it firs… - [One model is a single point of failure: routing between hosted and local AI models](https://larsima.com/hr/insights/routing-between-ai-models) — 2026-09-30 — Wiring a feature to a single model provider looks like the simple choice — until that provider gets slow, changes price, or fails on one kind of input. Routing across models isn't a cleverness upgrad… - [Designing for failure: real-time software that assumes the network will drop](https://larsima.com/hr/insights/real-time-assumes-the-network-drops) — 2026-09-30 — Most real-time systems are designed for the happy path and treat network loss as the exception. In the live-call product we built, disconnection is the rule — and that assumption reshapes reconnectio… - [Privacy by architecture: the data you never collect can't leak](https://larsima.com/hr/insights/privacy-by-architecture) — 2026-09-30 — The best privacy policy isn't code that deletes data — it's a schema that never made room for it. Data minimization is an architectural decision, not a legal clause. - [Why the on-call rotation matters more than the initial architecture](https://larsima.com/hr/insights/on-call-shapes-architecture) — 2026-09-30 — A system "designed to scale" and a system "designed to be woken up for" are not the same thing. The difference is only visible to whoever actually answers the page. - [Observability that isn't decoration: dashboards people actually open during an incident](https://larsima.com/hr/insights/observability-that-is-not-decoration) — 2026-09-30 — Every company has a metrics store and a stack of dashboards. The real question isn't whether you have observability — it's which screen you actually open during a real incident. - [Live speech translation: where model speed fights speech quality](https://larsima.com/hr/insights/live-speech-translation-latency) — 2026-09-30 — An ASR-to-MT-to-TTS pipeline adds latency at every stage. In the live multilingual meetings product we built, the real decision wasn't how to make latency zero, but where in the pipeline latency was… - [Why evaluating AI output isn't like testing software](https://larsima.com/hr/insights/evaluating-ai-output-is-not-testing) — 2026-09-30 — Traditional tests have one correct answer and an assert that checks it. Evaluating a language model's output means judging a range of acceptable answers that keeps shifting underneath you — and that… - [Documenting an API for AI agents is not documenting it for humans](https://larsima.com/hr/insights/documenting-apis-for-ai-agents) — 2026-09-30 — Ambiguity a human developer can shrug off and route around is, for an agent building a tool call, a generation error. The two audiences need different documentation. - [Cross-platform vs. native mobile: an engineering decision, not a marketing one](https://larsima.com/hr/insights/cross-platform-vs-native-mobile) — 2026-09-30 — The "one codebase" pitch hides where native actually breaks even. The real question isn't which framework to pick, but which subsystems in your specific product force a native escape hatch. - [The p99 that was a cron job](https://larsima.com/hr/insights/p99-was-a-cron-job) — 2026-07-28 — For three weeks one endpoint’s p99 went to nine seconds every night at 02:10 and recovered on its own by 02:40. Nobody was awake to see it and the dashboard averaged it away. The fix was one word of… - [What actually connects: TURN, and the networks that fight it](https://larsima.com/hr/insights/turn-relays-under-restrictive-networks) — 2026-06-11 — Every WebRTC tutorial ends at “and then ICE finds a path”. On the networks our users are actually on, it often does not. Three years of running a meetings product here, and the connectivity numbers t… - [SQLite in production, on purpose](https://larsima.com/hr/insights/sqlite-in-production-on-purpose) — 2026-04-30 — We run several real sites on a single SQLite file, and we would do it again — but not for the reasons people usually give, and not without the four things that make it survivable. Including the one t… --- Source: https://larsima.com/hr/insights Organisation: LARSIMA (شرکت فن‌آوران توسعه لار سیما), registered in Iran, no. 318515, since 2007-12-02.