← Back to all posts
8 min readCentrali Team

Centrali 2.4.0: AI Everywhere

Our biggest release yet brings AI-powered schema generation, natural language search, intelligent data validation, anomaly detection, and a new AI microservice.

ProductAnnouncementChangelogai

Centrali 2.4.0 is here, and it's our most ambitious release yet. We've infused AI throughout the platform to make building applications faster, smarter, and more intuitive.

This release introduces six major AI-powered features that transform how you work with data:

AI Schema Generator

Describe your data in plain English, and Centrali generates a complete schema for you.

How it works:

Click "Generate with AI" when creating a new structure, then describe what you're building:

I need a structure for customer orders. Each order has a customer name,
email, list of line items with product name and quantity, a total amount,
order status (pending, processing, shipped, delivered), and timestamps
for when it was created and last updated.

Centrali's AI analyzes your description and generates:

  • Field names and types (string, number, boolean, array, object)
  • Validation rules (required fields, formats)
  • Appropriate render settings for the console UI
  • Sensible defaults based on common patterns

You can review and edit the generated schema before saving. No more manually configuring each field.

Natural Language Search

Search your data the way you think about it.

Instead of building complex filters, just type what you're looking for:

orders from last week with status shipped
customers in Lagos with more than 5 orders
products under $50 that are in stock

The AI parses your query and translates it into the appropriate filters, operators, and structure targets. Results appear instantly.

In the Console:

Press Cmd+K and type your query naturally. The AI understands context, dates, comparisons, and relationships.

Via API:

typescript
const results = await centrali.search('high-value customers who signed up this month', { mode: 'natural' });

Natural Language Record Creation

Create records by describing them instead of filling out forms.

Click "Create with AI" on any structure and describe the record:

New order for John Smith (john@example.com),
3x Blue Widget at $29.99 each and 1x Premium Package at $149.99,
total $239.96, status is pending

The AI extracts fields, validates them against your schema, and shows you a preview. Edit any field before saving, or create the record directly.

This is especially powerful for:

  • Data entry - Type naturally instead of clicking through forms
  • Migration - Describe records from other systems
  • Testing - Quickly create test data with realistic values

Compute AI Assistant

Generate function code from descriptions.

When creating a compute function, describe what you want it to do:

When an order is created, calculate the total from line items,
apply a 10% discount if total exceeds $100, update the order record,
and send an email confirmation using the order-confirmation template

The AI generates working JavaScript code that:

  • Uses the correct Centrali APIs (api.fetchRecord, api.updateRecord, etc.)
  • Handles errors appropriately
  • Follows best practices for compute functions

Review the generated code, make any adjustments, and deploy. No more starting from a blank editor.

AI Data Validation & Cleaning

Let AI catch data quality issues before they become problems.

Real-Time Validation

Enable AI validation on any structure, and Centrali analyzes records as they're created or updated:

  • Format issues - Phone numbers, emails, dates in wrong formats
  • Inconsistencies - "New York" vs "NY" vs "new york"
  • Missing data - Required fields that should have values based on other fields
  • Outliers - Values that seem wrong based on the data pattern

Batch Scanning

Run a validation scan across all existing records to find and fix historical issues:

  1. Go to your structure's AI Validation tab
  2. Click Scan Records
  3. Review suggestions grouped by issue type
  4. Accept individual fixes or apply all

Auto-Correction

For trusted patterns, enable auto-correction to fix issues automatically:

Auto-corrected: "john.smith@gmial.com" → "john.smith@gmail.com"
Auto-corrected: "united states" → "United States"
Auto-corrected: Phone "1234567890" → "+1 (123) 456-7890"

You control which fields allow auto-correction and which require manual review.

AI Anomaly Detection & Insights

Centrali now continuously analyzes your data to surface insights and catch anomalies.

Pattern Detection

The AI learns your data patterns and alerts you when something unusual happens:

  • Unusual values - An order total of $0 when the average is $150
  • Missing references - Records pointing to deleted or non-existent records
  • Frequency changes - Suddenly receiving 10x more records than usual
  • Data drift - Field values shifting over time

Insights Dashboard

The new AI Insights tab on each structure shows:

  • Data quality score - Overall health of your data
  • Top issues - Most common problems to fix
  • Trends - How your data is changing over time
  • Recommendations - Suggested schema improvements

Reference Expansion

When analyzing anomalies, the AI automatically expands reference fields to give you full context:

Anomaly: Order #1234 has unusually high discount (85%)
Customer: Acme Corp (customer since 2023, 47 previous orders)
Recommendation: Verify discount is intentional for this VIP customer

Auto-Schema Discovery

For scenarios where you don't know your schema upfront, enable schemaless mode or auto-evolving mode:

Schemaless Mode

Accept any JSON data without predefined fields. Perfect for:

  • Ingesting data from external APIs
  • Prototyping before finalizing your schema
  • Storing arbitrary metadata

Auto-Evolving Mode

Start with a basic schema, and Centrali suggests new fields as it sees new data:

  1. You define initial fields (e.g., name, email)
  2. A record arrives with phone and company fields
  3. Centrali suggests adding these fields to your schema
  4. Accept or reject each suggestion

This bridges the gap between strict schemas and complete flexibility.

New AI Microservice

All AI features are powered by a new dedicated AI microservice that handles:

  • Natural language processing and parsing
  • Schema inference and generation
  • Validation rule learning
  • Anomaly detection models

The service runs asynchronously via a job queue, so AI operations don't block your main workflows. Results stream back in real-time as they're ready.

Other Improvements

  • Console UI polish - Unified favicon, smoother animations
  • Notification improvements - Better WebSocket reconnection and routing
  • Subscribe to notifications - One-click subscription to schema discovery events

What's Next

We're continuing to expand AI capabilities:

  • AI-powered smart queries - Generate complex queries from descriptions
  • Predictive validation - Suggest values before you type them
  • Cross-structure insights - Find patterns across related data

Get Started

Update your SDK to the latest version:

bash
npm install @centrali-io/centrali-sdk@latest

AI features are available now in the console. Visit docs.centrali.io for detailed guides on each feature.

Check the changelog for the complete list of changes.

Building something with Centrali and want to share feedback about this feature?

Email feedback@centrali.io