01
What Evidence an AI Feature Needs Before You Ship It
Ship readiness for an AI feature is one artifact per failure class, each with a stated minimum bar and a stated way it can be invalidated. Built here with a runnable 300 item fixture.
Reliable AI Systems in Production
Evals, retrieval, tools, and observability
A software-engineering view of AI-enabled products: where models help, where deterministic boundaries remain necessary, and how to operate the whole system responsibly.
The model behind the series
A useful AI feature is still a production system. It depends on permissions, identity, current data, tool contracts, latency budgets, evaluation, and a recovery path when a probabilistic component behaves differently than expected.
This series focuses on those engineering boundaries. The model is important, but it is one component inside a system that still needs clear invariants and evidence that it works.
Core principles
01
Keep deterministic boundaries explicit
Use models for useful ambiguity, while permissions, state transitions, money, and irreversible actions remain enforceable in code.
02
Evaluate the real workflow
Measure retrieval, tool selection, outputs, latency, and cost against representative tasks instead of judging a demo by feel.
03
Make failure inspectable
Record the inputs, versions, sources, tool calls, and decisions needed to explain and improve a production result.
Start here
Entry points into this subject, ordered as a reading path.
02
The Final Answer Can Be Right While the Tool Trajectory Is Unsafe
A trajectory grader for detecting unauthorized reads, duplicate effects, fabricated tool results, waste, and unsafe recovery even when an AI agent's final answer looks correct.
03
Eval Scores Move Between Runs: How to Compare Two Versions Honestly
A simulation of a real 5 point improvement showing how often a fixed eval threshold gives a random red build, and why a paired comparison detects the gain at a much smaller sample size.
04
Make Retried Agent Actions Idempotent Before Adding Autonomy
Agent retries must reuse one logical action identity and reconcile uncertain effects. An effect ledger makes autonomy bounded, inspectable, and recoverable.
Articles
Through the layers
Carry User Authorization Through Every AI Tool Call
A reference flow for keeping user identity, tenant, audience, and allowed actions attached to every AI tool call without trusting the model to enforce permissions.
Through the layers
Validate Model-Generated Tool Arguments at the Boundary
Structured model output is not trusted application input. Validate syntax, semantics, authorization, and effect policy again at the tool execution boundary.
Measurement
A Record-and-Replay Harness for Agent Tests
A working record-and-replay harness for agent tests: a JSON cassette of decisions and tool results, deterministic replay with zero live calls, and a readable diff when the trajectory changes.
Measurement
Calibrating an LLM Judge Against Human Disagreement
A practical agreement experiment for measuring an LLM judge against human labels, including ambiguous cases, position bias, confidence, and release thresholds.
Measurement
Evaluate Retrieval Separately From Answer Generation
A retrieval benchmark for separating missing evidence from generation failures using Recall@k, Precision@k, MRR, unanswerable questions, and controlled oracle context.
Measurement
Regression Testing Across Prompt and Model Changes
A versioned evaluation matrix for separating prompt, model, retrieval, tool, and grader changes while protecting safety, quality, latency, and cost.
Measurement
Golden Tests for Non-Deterministic AI Outputs
A graded assertion approach for regression-testing variable AI outputs without comparing every generated word to one frozen answer.
Measurement
Offline Evals Predict; Online Signals Correct
A measurement architecture that joins repeatable offline evals with production outcomes, safety signals, canaries, and privacy-aware review.
Measurement
A Failure Taxonomy for AI Features That Teams Can Actually Measure
A practical AI failure taxonomy that separates what happened, where it happened, its impact, and the suspected cause so teams can build useful metrics.
Measurement
Sampling AI Outputs for Human Review Without Only Seeing Easy Cases
A purely random sample hides rare and expensive AI failures. Combine baseline, risk-triggered, and exploratory samples without corrupting the denominator.
Measurement
Red-Team the Workflow, Not Only the Prompt
A strong system prompt cannot secure retrieval, tools, authorization, retries, or side effects. Red-team the complete AI workflow and its recovery paths.
Measurement
Turn Production Traces Into an Evaluation Set Without Copying User Data
A practical pipeline for learning from production AI failures while minimizing personal data, preserving the failure signal, and keeping evaluation fixtures reviewable.
Shorter pieces and reference
Older, introductory or narrower pieces on the same subject.
- Why AI Systems Still Depend on Ordinary Data Engineering
- Give Every Agent Tool Call a Deadline
- How AI Is Eating Your Codebase From the Inside
- How to Build a Secure Internal AI Tool
- RAG vs AI Agents vs Workflow Automation: What Should a Startup Build First in 2026?
- What Is RAG? A Practical Explanation for Startup Founders
- When You Do Not Need an AI Agent
- Understanding Model Context Protocol (MCP) for Developers
- Vector Embeddings for Code Search and Similarity
Engineering context
AI capability with ordinary engineering discipline
I help teams turn AI-enabled workflows into maintainable products with clear data, evaluation, permission, and failure boundaries.