What's shipped, what's next, and the reasoning behind the calls made along the way.
Paste PostgreSQL, MySQL, SQL Server, or SQLite DDL and get an interactive diagram: table nodes, PK/FK badges, auto-layout, pan/zoom, and a minimap. Parsing runs entirely in your browser — the DDL never touches a server unless you run an AI feature.
Deterministic checks for missing FK indexes, orphan tables, and relationships inferred from naming conventions but never formally declared — shown as dashed amber edges next to solid green ones, each ranked by severity.
Projects → Collections → Saved DDLs, an append-only version history per schema, a Draft/Ready status pill, and a dashboard queue of ready-for-review schemas with diff summaries and AI-drafted Change Requests.
Two Gemini-backed modes: AI Engineer finds and fixes problems in an existing schema as a selectable checklist; AI Designer proposes new tables and columns from a plain-language description. Both stage changes as a diff you accept or discard — nothing lands unreviewed.
Stripe checkout, a customer portal, and a monthly Gemini token cap (50K Free / 1.5M Pro) — the only difference between tiers. The canvas, gap detection, projects, version history, and the review workflow are unlimited on both.
17 articles across 6 categories covering the whole workflow, no sign-in required.
Point at a real Postgres or MySQL database and get a schema instantly, instead of hand-pasting a DDL snapshot. Read-only, SSL-required, and the connection string is never stored.
A GitHub App that runs gap detection against a schema file on every pull request and posts the diff summary as a comment, linking back into the full diagram.
Forward and rollback ALTER TABLE statements generated from a schema's diff, sitting next to the existing Change Request text instead of stopping at prose.
A read-only URL for any schema, embeddable in a README or a Notion page. Every schema is strictly private today.
A lightweight bug/feature-request form inside the app — right now the only channel is replying to a transactional email, which nobody does.
Generate a Prisma schema, Drizzle schema, SQLAlchemy models, TypeScript interfaces, or Zod/Pydantic validators straight from the parsed schema, no AI required.
Natural-language-to-SQL, grounded in the actual tables, columns, and relationships already on the canvas.
A members list per project — every project is single-owner today. The foundation a future team plan would sit on.
Flag columns that look like personal data and missing constraints around them, as a new category alongside the existing structural and AI-found issues.
Notes from building this — the reasoning, not just the changelog.
Password reset has its own flow, but there's still no way to verify your email, change it, or delete your account from inside the app. Reach out if you need an account removed in the meantime.
A standalone /account page ended up feeling like a second, half-empty dashboard sitting next to the real one. Its content — account info, subscription, AI usage — now lives directly on the Overview page, and /account just redirects there.
The side-by-side DDL diff tool wasn't landing well in practice. Rather than iterate on it indefinitely, it was pulled entirely, with no replacement planned right now.
The diagram canvas, gap detection, projects, version history, and the Draft/Ready workflow are unlimited on both tiers. Storage and workflow features don't cost enough at Free-tier scale to justify gating them, and feature-stripping against free ERD tools would hurt adoption more than it protects revenue.
A one-table schema and a fifty-table schema cost wildly different amounts to analyze. Tracking exact Gemini tokens (50K Free / 1.5M Pro monthly) instead of a flat number of runs means the cap actually tracks cost instead of penalizing bigger schemas.
Paste a schema and see your first analysis in under ten seconds.
Schema visualization & analysis for people who ship databases.