Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: Observability
Taking Agents to Production With Foundry Agent Service and Agent Framework
How a sales-prep agent moves from local code to a hosted, identity-scoped, sandboxed and monitored deployment, published in Microsoft 365 and grounded in business data.
11 min readAgent Harness Engineering: State, Verification and Control Around the Model
A runtime view of production agents: why prompts and context aren't enough, and how durable state, enforcement, verification and observability close the gap.
13 min readAI Architecture for Production Agents: Failure Modes and Control Layers
A layer-by-layer architecture guide for production agent systems: failure modes, guardrails, evals, observability, cost control, zero trust and governance.
12 min readWhen the Dashboard Is Green and the Agent Is Wrong: Tracing State
The dashboard is green and the agent is wrong: failures hide in state passed between healthy steps. What tracing, routing records and evals must capture.
12 min readSpring AI in the Enterprise: A Support Agent with Tools, RAG and Review
Spring AI for an incident-triage assistant: prompts as contracts, structured output, tool calling, MCP discovery, Qdrant RAG, tracing and bounded agent loops.
13 min readSelf-Improving Agents: Healing, Learning and Controlled Change
Self-improving agents without the danger: runtime self-healing, learning across runs, and changes that ship only after evals and canaries, on LangGraph.
15 min readAI Agent Audit Logs Need 12 Fields: What LangChain, CrewAI and Four Others Record
A static count of six agent frameworks, from LangChain and CrewAI to the OpenAI Agents SDK, against a 12-field audit trail draft: median coverage is 5 of 12.
10 min readWho Grades the Grader? Validating Agent Eval Suites Before You Trust Them
Who validates the eval suite? Judge drift, outcome-only scoring, private failure modes and hidden cost can make LLM evals lie. Checks for the suite itself.
7 min readWhy Production AI Agents Need More Than a Good Prompt
Production AI agents need more than a good prompt: context, tools, memory, failure handling, observability, testing and autonomy matched to the actual risk.
7 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.