Archives

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.

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: missing decision provenance, not metrics. The 3 observability gaps traditional monitoring won't catch.

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.

Layer AI agents over legacy systems without modernization. 30-80% productivity gains in 3-6 months. Patterns that bypass technical debt.

Slopsquatting: attackers register packages AI hallucinates. XZ Utils showed the stakes. A framework to assess your AI supply chain exposure.

50% cost reduction with subagent architecture for AI coding. Capable models for planning, fast models for building. Real metrics from Goose.

AI code review in CI/CD without prompt injection. Defensive patterns: three security tiers, isolated execution, no secrets in prompts.

From typing code to evaluating proposals. 70-80% time savings when AI explores options, you make the call. Real metrics from production.

GitHub Pages to Cloudflare Pages in 2 hours with zero downtime. AI-assisted DNS, hosting, and CI/CD migration with validation and real metrics.