Blog
Guides for theAI SQL workforce.
What we publish here: how AI analysts work, how SQL agents stay verified, how semantic layers ground them, and how skills travel between Claude, Codex, and Cursor.
All posts
ai analytics
What Is an AI Analyst? Skills & Tools
An honest guide to the AI analyst: how it turns plain-English questions into verified, read-only SQL on PostgreSQL, and where it still needs a human.
ai sql
The AI SQL Workforce: Skills, Not Models
An AI SQL workforce runs your verified queries as compiled, reusable skills. Why the skill library — not the model — is the asset.
agent skills
Claude Code Skills for SQL: SKILL.md Guide
How Claude Code Skills work for SQL: SKILL.md frontmatter, triggers, mutability, and compiling personas from verified Postgres queries.
ai agent configuration
CLAUDE.md vs CHION.md for SQL
CLAUDE.md defines agent behavior. CHION.md packages verified SQL skills your team already trusts. How the two compose into one portable workflow.
ai sql
Portable SQL Skills, Any AI Agent
Portable SQL skills compile your team's queries into a SKILL.md bundle that runs verbatim across Claude Code, Codex, Cursor, and Chion with no lock-in.
data architecture
Semantic Layer for SQL, Explained
A semantic layer for SQL maps business terms to the right table, column, and filter, the difference between a verified answer and a confident hallucination.
ai sql
SQL Agent: Read-Only, Verified, Auditable
A SQL agent runs queries on your behalf. The four properties to demand before trusting one: read-only, vaulted credentials, audit trail, verified skills.
ai sql
How Text-to-SQL Works on PostgreSQL
Text-to-SQL looks like one box on a slide. In production it's five distinct stages on PostgreSQL. Here's how the pipeline works and where it still breaks.