Orchestrating AI Agents: A Subagent Architecture
50% cost reduction with subagent architecture for AI coding. Capable models for planning, fast models for building. Real metrics from Goose.
Hugues Clouâtre - AI & Platform Engineering
AI and platform engineering insights from an AWS veteran and technology executive. Practical guides for CTOs and engineering leaders.
Latest posts
Agent-Ready Websites: The 5-Layer Implementation Stack
Your website serves two audiences now: humans reading HTML and agents calling tools. Measure agent readiness across five layers and learn what to ship first.
AI SDLC Governance: Three Layers for Engineering Leaders
A three-layer governance stack (comprehension, review gate, and observability) gives engineering leaders measurable control over AI-generated code volume.
How Does Context Engineering Drive Multi-Agent Reliability?
Context engineering turns AI context into infrastructure, preventing the accumulation, starvation, and leakage failures that undermine multi-agent reliability.
AI-Augmented CI/CD: Shift Left Security Without Risk
AI code review in CI/CD without prompt injection. Defensive patterns: three security tiers, isolated execution, no secrets in prompts.
AI-Assisted Development: The Accountability Layer
AI made writing code faster. Without governance controls, it makes reviewing and understanding code slower. The accountability layer is the structural response.
Open-Weight LLMs Reach the Structured Output Quality Ceiling
Open-weight models now match closed-source on structured output at 95x lower cost. Pre-registered blind eval, 30 samples, zero quality delta.
AI Approval Gates: Engineering Oversight at Machine Speed
AI approval gates: reversibility-tiered human-in-the-loop design with four health metrics that prevent reviewer atrophy at machine speed.
AI Agents in Legacy Systems: ROI Without Modernization
Layer AI agents over legacy systems without modernization. 30-80% productivity gains in 3-6 months. Patterns that bypass technical debt.
What a Null Result Taught Us About AI Agent Evaluation
We tested prompt repetition on 20 parallel AI agents. Ceiling effects dominated both experiments. The null result is a finding about evaluation design.
Why Your AI Agent Failed in Production
Why your AI agent failed: missing decision provenance, not metrics. The 3 observability gaps traditional monitoring won't catch.
RAG for Legacy Systems: 7,432 Pages to 3s Answers
7,432 pages to 3-second answers. Production RAG for legacy systems with model-agnostic reranking. No vendor lock-in, validated across 4 LLM families.
SRE for AI Agents: Error Budgets, Trust, and 90 Trials
Can an AI agent predict scope without hallucinating? We ran 90 trials. It added 1.7 phantom files per change. Error budgets and trust ladders are the gate.
AI Adoption in Engineering: Breaking the 50% Plateau
Purpose-built AI tooling cuts per-task cost 21-68%. Three-cohort model and four-phase operating framework for engineering leaders past the 50% adoption plateau.
AI-Governed DNS Migration Without Maintenance Windows
Programmatic pre-validation eliminates the weekend window from DNS migrations. Full platform migration: 2 hours, business hours, zero downtime.
AI Delivery Decision Frameworks: Type 1, Type 2, DACI
Misclassifying reversible decisions costs more than the decision itself. Four frameworks unblock AI delivery: Type 1/Type 2, Eisenhower, DACI, and PMBOK.
AI Supply Chain Attacks: New Vectors in Dependencies
Slopsquatting: attackers register packages AI hallucinates. XZ Utils showed the stakes. A framework to assess your AI supply chain exposure.