Nikolai Mazurov: blogNotes from a practitioner: AI, architecture, multi-agent systems.https://mazurovn.ru/Agent Capability Supply Chainhttps://mazurovn.ru/en/blog/agent-capability-supply-chain/https://mazurovn.ru/en/blog/agent-capability-supply-chain/Agentic systems gain a new architectural layer: capability becomes a versioned, verifiable, governed artifact — almost like a software component.Tue, 01 Sep 2026 00:00:00 GMTAgent systems are moving from loop-centric to contract-governed architecturehttps://mazurovn.ru/en/blog/agent-systems-perehodat-ot-loop-centric-k-contra/https://mazurovn.ru/en/blog/agent-systems-perehodat-ot-loop-centric-k-contra/The first AI agents were built around a simple loop. Once an agent runs long, changes system state and spends resources, a managed architecture grows around the model: intent, capabilities, policy, evidence, validation.Mon, 31 Aug 2026 00:00:00 GMTIt's not the model that decides, but the environment around ithttps://mazurovn.ru/en/blog/agent-harness/https://mazurovn.ru/en/blog/agent-harness/A year ago, an agent was assembled from a model and a list of tools. Now a separate layer has grown up around the model, and it is this layer that determines whether the system works or falls apart on the third step.Wed, 26 Aug 2026 00:00:00 GMTAI in business processes: where it pays, where it's hypehttps://mazurovn.ru/en/blog/ai-in-business-processes/https://mazurovn.ru/en/blog/ai-in-business-processes/Documents, support, monitoring: AI pays back fast there. Full autonomy without control: that is where the losses begin.Mon, 17 Aug 2026 00:00:00 GMTSkills, agents and the future of engineeringhttps://mazurovn.ru/en/blog/skills-and-agents/https://mazurovn.ru/en/blog/skills-and-agents/Expertise gets packaged into skills, routine goes to agents. What is left for the engineer, and what is worth learning right now.Mon, 03 Aug 2026 00:00:00 GMTADR and SDD: documents that don't gather dusthttps://mazurovn.ru/en/blog/adr-and-sdd/https://mazurovn.ru/en/blog/adr-and-sdd/Short decision records and specs before code. Why they matter even more in the age of agents, and why fat specifications are dead.Mon, 20 Jul 2026 00:00:00 GMTWhere AI actually helps in medicinehttps://mazurovn.ru/en/blog/ai-in-medicine/https://mazurovn.ru/en/blog/ai-in-medicine/No fantasies about replacing doctors: routine work, transcription, a second pair of eyes on scans, and patient follow-up. And why responsibility stays with the human.Mon, 06 Jul 2026 00:00:00 GMTAI on the factory floor: don't start with the modelhttps://mazurovn.ru/en/blog/ai-in-production/https://mazurovn.ru/en/blog/ai-in-production/How to bring AI into production cycles without pain: process and data first, a pilot on a boring bottleneck, and only then model debates.Mon, 22 Jun 2026 00:00:00 GMTOne agent is fine, an orchestra is betterhttps://mazurovn.ru/en/blog/multi-agent-orchestration/https://mazurovn.ru/en/blog/multi-agent-orchestration/Why a zoo of ten agents never works, and what actually fixes it: an orchestrator, context boundaries and quality control.Mon, 08 Jun 2026 00:00:00 GMT