Skip to main content

Writing

Notes on prompt engineering, evaluation and routing

June 27, 2026Why AI Cover Letters All Sound the Same (And How to Fix It)"I am writing to express my interest" is the tell. Here's what hiring managers notice, why AI defaults to generic trait claims, and the six-item prompt that gets you a letter worth reading.ContentJune 27, 2026Why AI Product Descriptions All Sound the Same (And How to Fix It)"Introducing the [Product]!" is the tell. Here's why AI product copy defaults to feature lists with no buyer in mind — and the six-item prompt that gets descriptions that convert.ContentJune 27, 2026Why Your AI LinkedIn Posts Sound Like AI (And How to Fix It)"Excited to share" is the tell. Here's why AI social posts have a recognizable pattern — and the prompt changes that make them sound like you wrote them.ContentJune 27, 2026Why Your AI Emails Sound Generic (And What to Do About It)The problem isn't the AI model — it's the prompt. Here's why AI-written emails come out off-tone and generic, and the specific changes that fix it on the first draft.ContentJune 27, 2026AI Marketing Copy That Converts: Fix the Prompt, Fix the OutputMost AI marketing copy sounds generic because the prompt was generic. Here's how to structure your request so AI delivers on-brand copy worth publishing.ContentJune 27, 2026How to Get AI Reports Worth SharingAI-generated reports fail because the prompt doesn't specify audience, structure, or depth. Here's what to add to get reports that are actually ready to distribute.ContentJune 10, 2026What Context Engineering Actually Means: From Vague Goal to Production Agentic SOPContext engineering goes beyond prompts. Learn how stateful SOP generation, tool inventories, task graphs, and OWASP security audits turn a single goal description into a production-ready agentic workflow.Context EngineeringJune 10, 2026Prompt Version Control: How to Manage, Roll Back, and Govern AI Prompts Like CodeHuman-readable slugs, immutable snapshots, environment scoping, state machine lifecycle, and HMAC-signed webhooks — the full governance model for treating prompts as production code.Prompt EngineeringJune 10, 20263-Tier Prompt Routing: How Rules-Based Optimization Avoids Unnecessary LLM CallsA composite routing score sends simple prompts through sub-10ms rules-based optimization with zero LLM call, and reserves full LLM calls for prompts that genuinely justify the cost.AI InfrastructureMarch 22, 2026The Problem With Eyeballing Prompt Quality (And What to Do Instead)Most developers judge prompt quality by feel. Here's what semantic drift detection, assertion testing, and context-aware scoring actually give you — and why vibes fail in production.Prompt Engineering
showing 10 of 27