/Orbit Blogs
Notes on AI engineering, system design, and the craft of building software well — from GenAI and agentic systems to the ideas behind shipping things that last.
Read the blog posts or check the About page for more info.
Recent Posts
AI Agents for Beginners: The Loop, the Context, and When to Hand Work to a Subagent
What an AI agent is, how the agent loop works, why context is what you're really spending, and when to use subagents vs. skills in Claude Code and Codex.
Text-to-SQL in 2026: The Benchmarks Got Solved. Your Database Didn't.
Text-to-SQL agents now score 83% on BIRD and 96% on Spider 2.0-Snow. What the top systems actually do, and what's still missing for your database.
Context Engines Explained: And How to Build the Half That RAG Misses
Your coding agent starts every task like a new hire. What a context engine is, why it cuts agent cost, and how to build its structured query layer.
Jev Explained: Is It Just Another Classifier?
Jev claims to solve classification up to 200x faster. Is it just another classifier? Same output, but you describe the task in words, not training data.
Featured
Letting a Fast Decision Model Drive Mobile QA: Three Practical Examples
Most QA agents wait on an LLM for every tap. TypeSafe's Jev makes quick, structured choices in 70-500ms instead — three examples of the pattern in action.
Shift-Left Security for AI-Assisted Development
AI generates code, tests and agents from a prompt, but security built for human-speed cycles can't keep up. Five principles for shifting security left.
Stop Giving Users Half-Answers: How to Connect Structured Data with Unstructured Context
Why your AI needs both the what and the why, and how to build an agent that joins SQL facts with document context to give complete answers.
Chunkless RAG: Why Flattening a Document Loses the One Thing That Made It Easy to Read
Chunk-and-embed RAG throws away document structure. Why reasoning over the original tree of headings and sections often gets you a better answer.