Database Development Without ORM: SQL-First Approach
Exploring a pragmatic approach to database development without ORM using Kysely and Ignatius for better SQL workflows, TypeScript integration, and AI safety.
Stock photo for illustration only, not from the actual event
- Many developers experience frustrations with traditional ORM tooling limitations
- New approach leverages Kysely for typed queries and full SQL support
- Includes built-in AI agent safeguards to prevent production data disasters
- Ignatius tool handles data modeling and planning using markdown workflows
Many developers frequently express frustration with ORM tooling, particularly concerning database migrations. While the desire to work directly with SQL is widespread, existing solutions often fall short when addressing testing, data seeding, staging environments, and team onboarding processes.
A fresh approach addresses these pain points directly without introducing unnecessary layers of abstraction. Built on top of Kysely, the tooling allows developers to maintain typed queries while gaining robust support for stored procedures and table-valued parameters. Meanwhile, the JavaScript and TypeScript layer focuses on IO coordination and business logic, leaving schema management entirely to the database layer.
Stock photo for illustration only, not from the actual event
The platform also integrates dedicated safeguards for AI agent interactions, featuring built-in skills, Model Context Protocol (MCP) support, and per-agent role configurations. This structure prevents automated processes from executing destructive modifications on production databases while maintaining highly efficient development workflows.
Moving away from traditional ORMs toward an SQL-first workflow highlights a growing demand for precise control over database performance and queries. As AI-assisted coding tools become ubiquitous, solutions like Kysely allow engineering teams to preserve crucial type safety without sacrificing the fine-grained optimization capabilities of native SQL.
Additionally, a separate tool named Ignatius handles comprehensive data modeling and planning. It supports multiple schema description formats, including modern IDEF1X symbols, and facilitates iterative development. Its markdown-based workflow ensures developers can easily preserve architectural context both for themselves and their LLM assistants.
This methodology eliminates the cumbersome translation layer residing between SQL and JavaScript, enabling developers to:
- Interact with relational databases using direct and predictable SQL queries
- Maintain seamless TypeScript integration alongside modern development tooling
- Collaborate safely with AI assistants through explicit permission boundaries
Ultimately, this solution establishes a practical middle ground between raw SQL management and full ORM frameworks, catering to developers who prefer direct database interactions while still demanding modern workflow support.
Source: Dev.to
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment