Rohit Singh ยท Agentic AI SystemsEssays and posts on production agentic AI: multimodal RAG, self-healing infrastructure, and autonomous agents.https://rohitsingh.ai/A Second Check, Alwayshttps://rohitsingh.ai/essays/a-second-check-always/https://rohitsingh.ai/essays/a-second-check-always/One model doing the work and an independent mechanism checking it is the single most reliable pattern I know for production AI. Here is how to build it.Wed, 02 Sep 2026 00:00:00 GMTessaysllm-as-judgeevaluationreliabilityRefuse Over Guesshttps://rohitsingh.ai/essays/refuse-over-guess/https://rohitsingh.ai/essays/refuse-over-guess/Why the most important design decision in a production AI system is deciding when the model is not allowed to answer, and how to engineer that refusal.Wed, 02 Sep 2026 00:00:00 GMTessaysagentic-aireliabilityverificationKimi K3: The Architecture Was Decided By The Kernelhttps://rohitsingh.ai/blog/kimi-k3-architecture-decided-by-the-kernel/https://rohitsingh.ai/blog/kimi-k3-architecture-decided-by-the-kernel/Reading the Kimi K3 technical report from the seat of someone who builds agentic systems that have to run on Monday morning: the decisions were made by kernels, caches, and harnesses, not loss curves.Thu, 30 Jul 2026 00:00:00 GMTmodel-analysisagentic-aiinferenceevaluationSkills, Connectors, and Subagents: Decoding the Architecture Anthropic Just Made Standardhttps://rohitsingh.ai/blog/skills-connectors-subagents/https://rohitsingh.ai/blog/skills-connectors-subagents/Anthropic's agent templates for regulated industries matter less for the customer logos than for the orchestration topology underneath: lazy-loaded skills, governed connectors, and isolated subagents.Sat, 09 May 2026 00:00:00 GMTmodel-analysisagentic-aiarchitectureregulated-industriesThe Open-Source Agentic AI: Kimi K2.6https://rohitsingh.ai/blog/open-source-agentic-ai-kimi-k2-6/https://rohitsingh.ai/blog/open-source-agentic-ai-kimi-k2-6/An honest read of Moonshot's Kimi K2.6 from someone building production agentic systems: why the open weights matter more than the leaderboard, and where it still falls short.Wed, 22 Apr 2026 00:00:00 GMTmodel-analysisagentic-aiopen-sourceevaluationBuilding Smarter AI Benchmarkshttps://rohitsingh.ai/blog/building-smarter-ai-benchmarks/https://rohitsingh.ai/blog/building-smarter-ai-benchmarks/What two controversial papers, The Illusion of Thinking and its rebuttal, taught us about measuring machine reasoning: many AI failures are benchmark design failures.Thu, 19 Jun 2025 00:00:00 GMTevaluationbenchmarksreasoningHITL vs Full Automation: A Decision Frameworkhttps://rohitsingh.ai/blog/hitl-vs-full-automation/https://rohitsingh.ai/blog/hitl-vs-full-automation/When to keep humans in the loop and when to ship fully autonomous agentic systems.Wed, 20 Mar 2024 00:00:00 GMTagentic-aihitlautomationdecision-framework5 RAG Failure Modes in Productionhttps://rohitsingh.ai/blog/rag-failure-modes/https://rohitsingh.ai/blog/rag-failure-modes/Common ways RAG systems fail in week 2, and how to avoid them with evaluation and observability.Fri, 15 Mar 2024 00:00:00 GMTragproductionevaluationlangfuse