Automated Rapid Application Development

You build the beautiful experience.

ARAD builds the robust backend and APIs.

ARAD generates, secures and deploys the backend as a live Liferay module — no pipeline, no downtime. Your team writes only what makes the app unique.

Zero pipeline
🔒 Auto rollback
♻️ OSGi hot-swap
🗄️ DB backup built-in
🤖 MCP native

Opening a store shouldn't mean building a new mall every time.

Security, roles, workflow, documents, navigation — every enterprise app needs the same foundations. Your AI agent rebuilds them on every project, then debugs them. That repetitive work is where most of the cost goes.

The hidden costs of unstructured enterprise AI: token waste, debug overhead, technology stack paralysis and a fragmented application landscape

Where your AI tokens go

Most of an enterprise module is backend plumbing and controls. ARAD takes that part off your agent's plate.

Building with an AI agent alone AI writes, reviews and debugs everything
Building with ARAD AI writes only what's unique
  • Database layer
  • REST APIs
  • Validation
  • Security & audit
  • Approval workflow
  • Forms & admin screens
  • Deployment
  • Your UI & business rules

Illustrative. Proportions vary by project.

ARAD automates the heavy part

Everything that used to burn tokens is now generated for you.

ARAD automation coverage: app and module management, field and metadata definitions, validation logic, API generation, database layer, list, detail and form views, deployment automation, admin interfaces, bulk upload and download, and an MCP layer per module

Liferay is the mall. Your app is the store.

New to Liferay? It's the platform ARAD builds on: an open-source enterprise platform, written in Java, that provides the shared foundations every serious application needs. Your team fits out the store, not the whole building.

🛡️Governance & roles
🔐Security & single sign-on
🔄Workflow & approvals
📁Document library
🧭Site & navigation builder
🧱UI fragments & themes
🔎SEO & search
🌐Multilingual sites

Building a mall for every store isn't realistic

For an enterprise these are must-haves, not extras. Building them yourself means years of work and a permanent maintenance bill: the cost of running a mall, when all you wanted was a store.

Option 1
Build everything yourself
The cost Every application re-implements governance, security, workflow, documents, navigation and SEO. It is slow, expensive and inconsistent across teams.
Option 2
Start from an enterprise platform
The shortlist SharePoint, AEM, SAP, OpenText, Sitecore or Liferay. The platform provides the enterprise capabilities, and your team builds only what is specific to your business.

Why we chose Liferay

Of those platforms, one is open and built on a stack enterprise teams already know.

🔓
Open source

Read, extend and host the platform yourself, instead of being locked into a closed product.

Java, end to end

A mature Java stack your developers already know, with full APIs and a module system to extend it.

🧩
Made to be extended

ARAD generates standard Liferay modules, so everything stays inside the platform's own rules.

From requirement to live module in four steps

Describe what you need in plain English. ARAD handles the rest inside Liferay, with no DevOps and no waiting.

01
💬
Describe

Describe the requirement in plain language to any AI agent: Claude, Codex, Cursor or VS Code Copilot.

02
🧠
Model

The agent turns it into a structured entity model: fields, relations, workflow and UI settings.

03
⚙️
Generate

ARAD generates production-grade Java, compiles it in-process and packages it as an OSGi bundle.

04
🚀
Deploy

ARAD backs up and migrates the database, hot-swaps the bundle, and the module is live. The portal never restarts.

Changes are just as quick

Say a price field has to be added to an existing form. Here's the work each way.

10 steps
Traditional or AI-coded
01DB migration 02Entity class 03Service layer 04REST API 05Form UI 06Validation 07Tests 08Commit & CI 09Deploy 10Restart portal
2 steps
With ARAD
01💬 Ask your AI agent to add the field 02🚀 ARAD generates & hot-deploys it

One model in. A complete enterprise module out.

Describe the fields your app needs. ARAD turns them into everything below.

🔌REST APISecured and ready to call
🖥️Forms & listsList, search, form and detail views
🔄ApprovalsSubmit, approve, reject, recall
🗄️DatabaseTables created and migrated safely
🧩Entity ModelFields described by your AI agent
🕘Audit historyEvery version of every record kept
🌐MultilingualTranslatable fields stored per language
🔐Encrypted PIIConfidential fields encrypted in the database
🤖MCP toolsAI agents can read and test the module

Under the hood: six production artifacts

Standard Liferay and OSGi components your team already knows how to run.

🔌
REST Microservice

Full JAX-RS REST API registered as an OSGi @Component service via the Liferay JAX-RS whiteboard. Versioned, secured, immediately consumable.

@Component JAX-RS OSGI
📝
Forms

Validated input forms generated straight from the entity model as JavaScript (Backbone) forms, so they can be reused in other applications, not just Liferay.

Validation Backbone.js Reusable
🖥️
Liferay Portlets

List view, search and detail portlets, fully scaffolded as OSGi portlets and ready to drag onto any Liferay page.

List view Form Detail
🔄
Workflow

Kaleo workflow process definitions (approval routing, roles, notifications and transitions) generated and registered automatically.

Kaleo Approval Routing
🗄️
DB Schema

DDL migrations (CREATE TABLE, ALTER TABLE) applied with a full database snapshot before, and automatic restore on failure.

DDL Backup Rollback
📥
Data Migration Tools

Excel and CSV import and export generated for every module, so data moves from a legacy system into Liferay without writing migration scripts.

Excel CSV Import / Export

Enterprise controls: built in, not prompted

These are the features that bloat AI-written code and fail audits when they're missed. ARAD includes them in every module.

Control AI alone ARAD
🛡️Role-based access on every APIPrompt & testManual✓ Built in
🕘Full audit history of recordsPrompt & testManual✓ Built in
👤Created-by & modified-by trackingPrompt & testManual✓ Built in
🔄Approval workflow & task assignmentPrompt & testManual✓ Built in
🔐Encryption of confidential fieldsPrompt & testManual✓ Built in
🌐Multilingual dataPrompt & testManual✓ Built in
Input validationPrompt & testManual✓ Built in
📋Master data & dropdown lookupsPrompt & testManual✓ Built in
📥Excel / CSV import & export for every modulePrompt & testManual✓ Built in
🗄️DB backup & auto rollback on deployPrompt & testManual✓ Built in

Custom OSGi modules. Without the drawbacks.

The drawbacks people worry about come from hand-coding modules. ARAD generates them to Liferay's standards, so you keep the power without the pain.

Custom OSGi modules remain “a powerful and comprehensive way to extend Liferay’s functionalities.” — Liferay documentation
The concern
Needs scarce Liferay specialists
With ARAD Low-code generation with no Java to write. Services, APIs, forms and UI templates come from one place, so no deep Liferay expertise is needed to build or update modules.
The concern
Slow to build by hand
With ARAD One step generates the whole module: services, APIs, forms, validation, relationships and admin pages. No assembling Objects, Collections and Client Extensions separately.
The concern
Breaks on Liferay upgrades
With ARAD The ARAD team tracks Liferay's breaking changes. Regenerate every module for a new Liferay version with a single click.
The concern
Risky to deploy
With ARAD Modules are hot-deployed as standard OSGi bundles, with a database backup before every change and automatic rollback if anything fails.
📦
Your source code

Download the generated modules into your own Liferay workspace. No lock-in.

🔁
Upgrade-ready

One-click regeneration moves every module to a new Liferay version.

☁️
Where it runs

Liferay PaaS and self-hosted DXP. Liferay SaaS doesn't allow custom OSGi modules.

Where Liferay Objects fall short

Liferay Objects work well for simple, self-contained data. Enterprise applications that need any of these end up writing OSGi modules anyway.

What you need Liferay Objects ARAD modules
🗄️Use an external database Not supported. Objects can only store data outside Liferay through the Salesforce and SugarCRM connectors. Supported
🪟Use existing tables or database views Not supported. Publishing an object always creates its own new table. Supported
🔄Transactions across several entities Limited. Webhook and client-extension actions run outside Liferay's database transaction, and Groovy script actions aren't available on Liferay SaaS. Real multi-entity transactions need custom OSGi code. Built in
📦All-or-nothing batch saves Not supported. Batch APIs keep earlier items saved when a later item fails. Supported

Based on Liferay DXP documentation and source code, September 2026.

Every deploy is safe and reversible

Each ARAD deployment is wrapped in a safety envelope. Schema and bundle changes are always atomic and always reversible.

🗄️
DB Backup
Snapshot before any change
📐
Schema Δ
DDL applied to database
♻️
Bundle Swap
Old uninstalled, new installed
Verify
Bundle ACTIVE health check
Live
Zero portal downtime
⚠️ Failure at any step triggers automatic rollback: the bundle is restored, the database snapshot is reverted, and the portal is unaffected.

The deployment layer vibe coding was missing

AI can write enterprise code. ARAD makes sure it actually runs in production.

⛔ No deployment path
AI code that never reaches production
✓ MCP call → live OSGi bundle
⛔ Schema drift
DB and code permanently out of sync
✓ Atomic backup → DDL → verify
⛔ Pipeline complexity
Git, Jenkins, CI/CD just to deploy
✓ Zero pipeline, in-process compile
⛔ Portal downtime
Every deploy restarts the portal
✓ OSGi hot-swap, zero downtime
⛔ Untested AI code
Bad generations break production
✓ Auto rollback on bundle failure
⛔ Unstructured output
Free-form code, hard to govern
✓ MCP enforces typed entity model

Use the AI agent you already have

ARAD exposes MCP tools, so any MCP-compatible AI agent can describe, model and deploy without touching a terminal or pipeline.

C
Claude
Anthropic
MCP native
O
Codex
OpenAI
MCP native
Cur
Cursor
AI IDE
MCP native
VS
VS Code
Copilot
MCP native
M
MCP
Protocol
Open standard

A familiar API pattern in every module

Every module ARAD builds gets its own secured API, and each one uses the same standard operations. Developers and AI agents learn the pattern once and apply it in every app.

Modules in your apps
🎓 Training portal
🗓️ Leave management
🚚 Delivery tracking
📊 Progress reporting
Standard operations in each module's API
find findByQuery search countByQuery calculateByQuery create update createMultiple submit approve reject recall
What you gain
Frontends are built fast from the live schema
No new API conventions to learn for each project
Fewer integration bugs
AI agents test with real data and real roles

The complete list

Every benefit, grouped by who it helps most.

💰
Lower AI cost
For budget owners
  • Fewer tokens burned. The backend is generated, not written by AI.
  • Less debugging. Proven generated code avoids error loops.
  • AI focus stays sharp. The agent's context is spent on your real problem.
  • Predictable spend. Effort no longer grows with backend size.
Faster delivery
For project leads
  • Requirement to live module in one flow: describe, model, generate, deploy.
  • Changes in minutes. Add a field and hot-deploy it.
  • Ready-made screens. List, search, form, detail and admin views come with every module.
  • Easy legacy migration. Every module comes with Excel/CSV import and export, so legacy data moves into Liferay without custom scripts.
🏛️
Enterprise-ready by default
For compliance & security
  • Role-based access enforced on every service call.
  • Complete audit trail. History of every record, plus who created and changed it.
  • Data protection. Confidential fields are encrypted before they are saved.
  • Approval workflows with submit, approve, reject, recall and task assignment.
  • Multilingual content and data out of the box.
🛟
Safe deployment
For operations
  • No pipeline. No Git, Jenkins or CI/CD needed for modules.
  • Zero downtime. OSGi hot-swap, and the portal never restarts.
  • Backup first. The database is snapshotted before every schema change.
  • Automatic rollback. Any failure restores the bundle and the data.
🤖
Built for AI agents
For developers
  • MCP native. Works with Claude, Codex, Cursor and VS Code Copilot.
  • MCP tools for every module, so agents can read the schema and data.
  • Live testing. Agents check APIs with real data under real user roles.
  • Consistent APIs. Every module gets its own API, built on the same standard operations.
🧭
Governed & consistent
For IT leadership
  • A typed entity model. No free-form, ungoverned AI code.
  • No schema drift. Database, code and documents stay in sync.
  • One standard stack. Ends framework debates and app silos.
  • Easier long-term support. Every app is structured the same way.

Don't believe it? Fair enough.
See it for yourself

Our experts from the product engineering team will show a live demo of how it works and answer all your queries and all follow-up queries down the line.