Tech notes
This article outlines five major technical gaps developers must bridge to become architects: shifting from monolithic to distributed thinking, mastering system design over feature coding, evolving from solo work to team collaboration, aligning technology with business strategy, and adopting cloud-native DevOps practices.
This article categorizes four types of conflicts product managers face—resource, cognitive, boundary, and emotional—and provides a practical framework for resolution through shared interests, role-specific communication, emotional bank accounts, and six diagnostic questions to turn disputes into trade-offs without damaging long-term collaboration.
This article dissects production-grade agent intent recognition, explaining how to move beyond simple classification by defining structured TaskSpecs, resolving multi-turn context, implementing hierarchical routing, calibrating confidence with risk-aware thresholds, and evaluating with metrics that prioritize correct routing over raw accuracy.
This article analyzes 10 production-verified ops agent systems (2024-2026), distills 10 architectural principles, compares MCP direct-connection vs UModel semantic-layer architectures, details ReAct loop boundaries with LangGraph implementation, and proposes a three-layer memory architecture combining Markdown runbooks with vector databases for scalable SRE automation.
This article details the integration of Hermes Agent with Karpathy's LLM Wiki, explaining a three-layer architecture (raw sources, AI-generated wiki pages, schema config) and three core operations (ingest, query, check) that turn fragmented notes into a continuously compounding, self-updating knowledge graph using YAML frontmatter and wikilinks within Obsidian.
This article analyzes the Dsh team's Super-agent design, showing how classic systems engineering — cache locality, fast/slow paths, speculative execution, and compute-storage separation — enables AI agents to maintain continuous context, handle concurrent events efficiently, act proactively on environmental triggers, and persist across device failures.
The article presents a Domain Facade design pattern for MCP servers that consolidates 60+ fine-grained tools into 9 domain-level tools, reducing per-interaction token consumption from ~12,000 to ~2,000 (85% reduction) while preserving full functionality, and introduces a manifest-based roundtripping pattern for atomic multi-file edits.
The article argues that the Harness — the agent runtime framework — is the true product for production AI agents, detailing 12 architectural principles covering task-oriented tool discovery, layered memory, credential isolation, approval workflows, code execution sandboxes, multi-channel adapters, state persistence, and system evaluation.
This article evaluates eight AI-powered and foundational operations tools for 2026 across alerting, logging, metrics, root cause analysis, and self-healing, providing applicable scenarios, limitations, and a team-size-based selection guide while emphasizing that solid observability foundations must precede AI adoption.
This article explains the four layers of ops monitoring (infrastructure, applications, middleware, business), details Google's four golden signals (latency, traffic, errors, saturation), and provides a step-by-step guide to implementing application monitoring with Spring Boot, Prometheus, and Grafana including alerting and continuous improvement practices.
