How to use Modej

Modej takes your database (Firebird, MySQL, PostgreSQL, Oracle, SQL Server, and others) together — optionally — with the source code that uses it, and gives you back:

• an interactive ER schema with all tables, columns, keys and relationships;
• a map of which parts of the code use which tables (and how: read, write, modify);
• automatic documentation generated by artificial intelligence;
• professional PDF reports ready to hand to a client or colleague.

Modej reads the DDL - the SQL script describing the database structure (tables, columns, keys, indexes, triggers, stored procedures). You never need a direct connection to your database: the text file with the schema is enough.

Access, dBase and Paradox don’t have a text DDL you can export directly — they are file-based, binary databases. For these you need a conversion script that runs on the computer where the database file lives, before uploading it to Modej.

Uploading your source code too (Delphi, C#, Java, PHP...) lets Modej understand which parts of the application use which tables — the basis for AI Analysis, Code↔DB, Process Map and full Technical Reports.

file.pas | UnitName | ClassName | SQL_STRING:SELECT * FROM TABLE WHERE...

Every table becomes a draggable card with all its columns, primary keys (PK) and foreign keys (FK) highlighted. Relationships are drawn automatically: indigo lines for real FKs declared in the database, orange lines for the manual FKs you add.

Group tables into coloured "areas" (e.g. Warehouse, Sales, Accounting) by dragging each table onto its area in the left sidebar. These colours show up everywhere: ER Schema, Glossary, Technical Report.

• Create new areas with a custom name and colour;
• Filter by "Assigned" / "Unassigned" to quickly complete the classification;
• Inherit colours from a previous version of the same database, if you’re updating an already classified schema.

Many legacy databases don’t declare all relationships at the SQL level — the relationship exists only "in the developers’ heads" or in application code. Manual FKs let you document them explicitly, and they show up in the ER Schema as dashed orange lines.

Select the source table and column, then the target table and column, add an optional note and save.

Upload a ZIP of the source code: Modej analyses it with AI algorithms and generates reports on 4 different fronts, selectable at the start:

Unlike AI Analysis (which gives a general overview), Deep Analysis examines every single file in the project, works out which business process it belongs to (e.g. "Order Management", "Customer Records") and automatically proposes a process map ready to apply.

When it’s done, click "Apply to Processes" to transfer the classification directly into the Process Map, saving the manual work of dragging.

Unlike AI Analysis (which looks at the code), AnalisiDB is deterministic — it doesn’t use AI: it applies fixed rules to keys, foreign keys, data types, naming, indexes and the normalisation level of the imported schema.

• An overall grade and an OK / Normal / Poor badge for the whole database;
• every issue found is grouped by table, with severity and a plain-language explanation;
• a corrective SQL script ready to copy — review and adapt it before running it, don’t fire it blindly at a production database.

For every imported code unit you see the list of tables it uses, with the operation type (SELECT, INSERT, UPDATE, DELETE, or "mixed" if it does more than one) and the exact method name. The statistics at the top summarise the overall picture: total relationships, how many SELECT/INSERT/UPDATE/DELETE, tables and units involved.

With the Query Extraction Plus you can export this whole matrix to CSV, JSON or SQL — useful for documentation, audits or migrations.

Create processes (e.g. "Invoicing") and sub-processes (e.g. "Issue Invoice"), then drag the still-unassigned code units from the left column into the right sub-process. Alternatively use "Analyse with AI" to have the entire structure proposed automatically based on the unit names.

Remember to click Save after every change — a green toast confirms the save succeeded.

Once the Process Map is populated, this page automatically generates an SVG diagram with the full hierarchy: processes at the top, sub-processes below them, and the code units assigned to each at the bottom.

Useful for a quick overview, or to export as an image for documentation.

See all imported versions for each database, with table count, import date and size. Rename a DB by double-clicking its name, delete outdated versions (if there’s more than one), or jump straight to the ER Schema or Comparison of each version.

Select a base version and a new version: you’ll see added tables (green), removed (red) and modified (orange) tables in a dual-view panel. Click a modified table to see exactly which columns changed, with the before/after type detail.

Similar to Version Comparison, but built for documentation: generates a changelog in "release notes" style with a visual timeline, exportable to .md to insert directly into the project’s Git repository.

Compares the database tables with those actually referenced in the imported code. "Unused" tables are likely legacy or obsolete — ideal candidates for cleanup before a migration, grouped by area with a percentage of the total.

List ranked from most to least used table, with coloured bars (blue=rare, green=moderate, orange=frequent, red=critical) and badges by operation type. Filter by SELECT/INSERT/UPDATE/DELETE for targeted analysis (e.g. DELETE only for a security audit).

For each table, write (or generate with AI) a description in business language, not technical — useful for anyone who needs to understand the database without being a developer. The "AI generate all" button automatically fills in every table still missing a description. Exportable to PDF.

Download all unit↔table relationships in CSV (for Excel), JSON (for other tools) or SQL (for direct queries). Perfect for technical documentation or to feed external analysis tools.

A 20-30 page PDF generated entirely in your browser (no data leaves your computer during generation), ready to hand to a client or archive:

Score from 0 to 100 calculated from the number of tables, manual FKs, average columns per table and a penalty for orphan tables above 20%. You also see the trend over time: if you’ve imported multiple versions of the same database, a chart shows whether the system is becoming more or less complex from version to version.

Turns the project data into fullscreen slides: cover, overview with 7 metrics (tables, columns, FKs, triggers, stored procedures, views, indexes), interactive ER schema, one slide per business process with its sub-processes and units, heatmap, and a final slide.

Navigate with the ← → arrow keys, exit with ESC. Perfect for client meetings or live demos.

Each Plus is purchased separately and is valid for 12 months from purchase. The catalogue below is read live from the marketplace, so it always matches what you actually find on the Plus page — names, prices and descriptions stay in Italian regardless of this guide’s language, for now.

Analisi

Analisi AI

Codice DB

Compliance

Migrazione

Report

Schema DB

Storage

Team

Features with a 🕐 Coming soon badge are on the way: advanced security audit, code version comparison, automatic documentation, dependency graph and more — the catalogue expands regularly.

The ☀/🌙 button at the top right of the main menu switches between dark and light theme. The preference saves automatically and applies to every page, including the ER Schema which lives in a separate panel — no need to set it twice.

Italian, English, Portuguese, Spanish, German, French and Kinyarwanda. Each user picks their own language from the selector in the main menu, independently of the other users on the same account — everyone sees Modej in the language they prefer.

AI-generated content (glossary, reports, descriptions) is created directly in the chosen language. This guide also follows the same language set in the menu.

• Passwords, API keys and credentials in the code are automatically masked before any submission to the AI;
• Source code is not kept on the servers after the analysis;
• DDL extraction scripts for legacy databases run locally on your computer — no credential is ever sent to Modej;
• See the Terms of Service and Privacy Policy for full details.

The account Owner can invite other users and assign each one of the three internal roles. Permissions are also checked server-side, not just by hiding buttons in the interface.

From Team the Owner sees all users, their status (active/disabled) and last access.

The base subscription (Monthly or Annual plan) is refundable within 14 days of purchase: just open a request from your profile, optionally state the reason, and the Modej team reviews it — once approved, the refund arrives via Stripe within 5-10 business days to the same payment method.

Permanently deletes a project and all linked data — database, code, processes, manual FKs. For safety you’ll need to type the exact project name to confirm. The operation is irreversible and transaction-protected: if something goes wrong halfway through, nothing is deleted.