Reduce AI token burn.
Let ARAD do the FAT part.
Your AI agent describes the app. ARAD generates, secures and deploys the backend as a live Liferay module, with no pipeline and no downtime. Your agent writes only what makes your app unique.
AI can write enterprise apps. The backend burns the tokens.
On every project, AI agents regenerate the same database layer, APIs, security and workflow, then debug it. That repetitive work is where most of the cost goes.
Where your AI tokens go
Most of an enterprise module is backend plumbing and controls. ARAD takes that part off your agent's plate.
- 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.
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.
Describe the requirement in plain language to any AI agent: Claude, Codex, Cursor or VS Code Copilot.
The agent turns it into a structured entity model: fields, relations, workflow and UI settings.
ARAD generates production-grade Java, compiles it in-process and packages it as an OSGi bundle.
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.
One model in. A complete enterprise module out.
Describe the fields your app needs. ARAD turns them into everything below.
Under the hood: four production artifacts
Standard Liferay and OSGi components your team already knows how to run.
Full JAX-RS REST API registered as an OSGi @Component service via the Liferay JAX-RS whiteboard. Versioned, secured, immediately consumable.
List view, search, form, and detail portlets, fully scaffolded as OSGi portlets and ready to drag onto any Liferay page.
Kaleo workflow process definitions (approval routing, roles, notifications and transitions) generated and registered automatically.
DDL migrations (CREATE TABLE, ALTER TABLE) applied with a full database snapshot before, and automatic restore on failure.
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.
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.
The deployment layer vibe coding was missing
AI can write enterprise code. ARAD makes sure it actually runs in production.
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.
One API language across every app
Every module ARAD builds, in any app, answers the same set of calls. Developers and AI agents learn them once.
Every benefit, at a glance
What ARAD changes for each part of your team.
REST API, portlets, workflow and database schema from a single entity model.
No Git, Jenkins or CI/CD needed to get a module live.
Modules are hot-swapped while users keep working.
Every module speaks the same calls. Learn once, reuse everywhere.
The complete list
Grouped by who it helps most.
- 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.
- 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.
- Quick legacy migration. Old tables map straight to new modules, with bulk upload for the data.
- 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.
- 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.
- 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.
- One API language. The same calls work across every app and module.
- 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.