Comparisons
OrmAI vs. the alternatives.
Honest, technical comparisons. We'll tell you the cases where the alternatives are actually fine.
All 4 follow the same shape. Each starts by describing the alternative on its own terms, works through the specific ways it comes apart once there is more than one tenant and more than one table, and names the conditions under which it is still the right answer — because for every one of these there are conditions where it is. A comparison that could not name them would not be worth reading.
Read the one that matches what you run today
vs. hand-rolled tools
OrmAI vs. hand-rolled tools
You have written one bespoke function per agent capability, and the list keeps growing. This one is about the six places the same work gets duplicated across tools, and when stopping to generalise is worth it.
Read the comparison →vs. MCP database servers
OrmAI vs. MCP database servers
A standalone MCP server sits in front of the database and any MCP-aware client can reach it. This one is about where that trust boundary sits, and how to keep MCP without giving up policy.
Read the comparison →vs. raw SQL
OrmAI vs. raw SQL
Your agent holds a tool that takes a SQL string and executes it. This one covers what that actually hands the model, the two failure modes it produces in production, and the narrow cases where it is still fine.
Read the comparison →vs. text-to-SQL
OrmAI vs. text-to-SQL
A model writes the SQL for you, and something downstream is supposed to validate it. This one is about why that validation step is harder than it looks, and what the empirical results say.
Read the comparison →Already decided?
Comparisons stop at the decision. The guides pick it up from there, starting with wiring the policy to the ORM you already run and ending with a checklist to review the result against before it sees production data. Definitions for the terms these pages lean on — tenant scoping, field-level redaction, query budgets, gated writes — are in the glossary.