Data Modeling

Entities, fields, and relationships.

Backdrift automatically designs your database schema from natural language. Describe your data, and we handle the rest.

Database Options

Backdrift supports two database backends:

Option	Best For	Features
DynamoDB (default)	Simple apps, high scale	NoSQL, auto-scaling, pay-per-request
RDS PostgreSQL	Complex queries, joins	Relational, SQL, foreign key constraints

To request PostgreSQL, mention it in your prompt:

"A CRM with users, contacts, and deals. Use PostgreSQL for complex reporting queries."

Entities

Entities are your data models. Each entity becomes a database table with automatic CRUD endpoints.

Defining Entities

"Tasks have a title, description, due date, priority, and status."

This generates:

interface Task {
  id: string;           // Auto-generated primary key
  title: string;
  description?: string;
  due_date?: string;    // ISO datetime
  priority: 'low' | 'medium' | 'high';
  status: 'pending' | 'active' | 'done';
  created_at: string;   // Auto-managed
  updated_at: string;   // Auto-managed
}

Field Types

Type	Description	Example
string	Text data	names, descriptions
number	Floating point	prices, scores
integer	Whole numbers	counts, ages
boolean	True/false	is_active, is_verified
datetime	ISO timestamp	due_date, created_at
date	Date only	birth_date
enum	Fixed options	status, priority
json	Arbitrary JSON	metadata, settings

Field Constraints

Required vs Optional

"Tasks have a required title and optional description."

Unique Fields

"Users have a unique email address."

Default Values

"Task status defaults to 'pending'."

Enums

"Priority can be low, medium, or high."

Relationships

One-to-Many

"Tasks belong to projects. A project has many tasks."

This creates:

• project_id foreign key on Task
• Index for efficient queries
• Cascade delete options

Many-to-Many

"Tasks can have multiple tags. Tags can be on multiple tasks."

This creates a join table automatically.

Self-Referential

"Tasks can have subtasks."

Creates a parent_task_id field on Task.

Automatic Fields

Backdrift adds useful fields automatically:

Field	Description
id	UUID primary key
created_at	When record was created
updated_at	When record was last modified
org_id	Tenant ID (if multi-tenant)

Indexes

Backdrift creates indexes for:

• Primary keys (id)
• Foreign keys (project_id)
• Tenant keys (org_id)
• Frequently queried fields

Query Optimization

"Users often search tasks by status and due date."

This hints Backdrift to create a composite index on (status, due_date).

Multi-Tenant Data

In multi-tenant mode, every entity gets:

interface Task {
  id: string;
  org_id: string;  // Auto-injected tenant ID
  // ... other fields
}

All queries are automatically filtered by org_id.

PII Detection

Backdrift detects Personally Identifiable Information:

Field Pattern	Detection
email	Email PII
phone	Phone PII
ssn	SSN PII (Critical)
address	Address PII

When PII is detected, we recommend:

• Encryption at rest
• Access logging
• Data retention policies

Example: E-commerce

"An e-commerce platform with:
- Products with name, description, price, and category
- Categories with name and description
- Orders with status, total, and customer
- Order items linking orders to products with quantity"

Generates:

┌─────────────┐     ┌─────────────┐
│  Category   │     │   Product   │
├─────────────┤     ├─────────────┤
│ id          │←────│ category_id │
│ name        │     │ name        │
│ description │     │ description │
└─────────────┘     │ price       │
                    └─────────────┘
                          ↑
┌─────────────┐     ┌─────────────┐
│   Order     │     │ Order Item  │
├─────────────┤     ├─────────────┤
│ id          │←────│ order_id    │
│ customer_id │     │ product_id  │─→
│ status      │     │ quantity    │
│ total       │     │ unit_price  │
└─────────────┘     └─────────────┘

Scheduled Jobs

Learn how to run background tasks on a schedule.