Field Craft Labs Blog https://fieldcraftlabs.com/blog Practical notes on data engineering, custom software, and AI enablement for small and mid-sized businesses, from Field Craft Labs in Chicago. en-us Thu, 23 Jul 2026 12:00:00 +0000 Text to SQL with LLMs, Part 4: Production-Grade Systems: Semantic Layers, Agents, Guardrails, and Evals https://fieldcraftlabs.com/blog/text-to-sql-production-semantic-layer-guardrails https://fieldcraftlabs.com/blog/text-to-sql-production-semantic-layer-guardrails Thu, 23 Jul 2026 12:00:00 +0000 How production text to SQL systems work: semantic layers, agentic self-correction, read-only guardrails, MCP integration, and the eval harness that proves accuracy. Text to SQL with LLMs, Part 3: Retrieval, Golden Queries, and Business Context https://fieldcraftlabs.com/blog/text-to-sql-rag-golden-queries https://fieldcraftlabs.com/blog/text-to-sql-rag-golden-queries Wed, 22 Jul 2026 12:00:00 +0000 Static schema context has a ceiling. Learn how retrieval, verified question-SQL pairs, a business glossary, and value grounding push text to SQL accuracy higher. Text to SQL with LLMs, Part 2: Schema Context Engineering That Actually Moves Accuracy https://fieldcraftlabs.com/blog/text-to-sql-schema-context-engineering https://fieldcraftlabs.com/blog/text-to-sql-schema-context-engineering Tue, 21 Jul 2026 12:00:00 +0000 The biggest accuracy lever in text to SQL is how you present your schema. Column descriptions, sample values, relationships, and curated views, with SQL to build them. Text to SQL with LLMs, Part 1: Why the Baseline Approach Fails on Real Databases https://fieldcraftlabs.com/blog/text-to-sql-llm-baseline-approach https://fieldcraftlabs.com/blog/text-to-sql-llm-baseline-approach Mon, 20 Jul 2026 12:00:00 +0000 Basic schema-in-prompt text to SQL still fails on real databases in 2026. See the current benchmark numbers, the five failure modes, and what actually fixes them.