Anatomy of a Prompt — System, User, and Assistant Explained
Understand how system, user and assistant messages shape LLM behavior.
The Harsh Tech
Latest articles
OpenAI Prompt Caching: Undocumented Cross-Model Behavior and Production Cost Implications
I'm building an AI agent from scratch—no frameworks, no abstractions—specifically to understand where every token goes and how much it costs. This is Phase 3 of my token economics research. Phase 1 covered basic tool calling mechanics. Phase 2 reveal...
Model Selection for AI Agents: Measuring Token Costs Across OpenAI's Model Family
I've been building an AI agent from scratch. No frameworks, no abstractions, to understand where every token goes and what drives cost at scale. In previous post/phase, I measured how tool definitions and conversation depth impact token usage. The fi...
Token Explosion in AI Agents: Why Your Costs Scale Exponentially
I built an AI agent from scratch. Not because frameworks aren't good. They are(and I suggest you use them). But because I needed to see where every token goes. When you're building production systems that could cost $150K+/year in LLM tokens alone, y...
OOP Fundamentals for AI Applications
Why Your AI Code Needs Better Structure
SOLID Principles for AI Systems: Why Your RAG Pipeline Needs Better Architecture
Your RAG pipeline works perfectly in staging. You deploy to production. 10,000 concurrent users hit it. Embeddings start timing out. Vector search fails silently. LLM calls retry infinitely because someone forgot to set a max. Your "AI-powered" featu...
Thread Wars: Episode 3 – Rise of the Virtual Threads
We finally got our threads back. Now let’s not burn the world down with them.