Application Development

From idea to deployed application — architect, build, test, preview, and ship entirely inside Opulent

What it is

Opulent is a full-stack development environment. The agent doesn't just write code snippets — it architects complete applications, builds them in a persistent sandboxed workspace, runs a live browser preview as it works, writes and runs tests, commits to GitHub, and deploys to production. You see the application render in the Browser panel as it's being built.

Quick Start

Build an Internal Tool

"Build a web dashboard that shows our MRR, churn, and top customers
from Stripe. Next.js, Tailwind, dark theme. I want to deploy it to Vercel."

Build a Full Application

"Build a B2B SaaS onboarding tracker. Users can:
- Create customer accounts
- Assign onboarding tasks with due dates
- Track completion percentage per account
- Send automated email reminders

Use Next.js, Convex for the database, Better Auth for login. Deploy-ready."

Prototype from Description

"Build a working prototype of our new pricing page — 3 tiers, feature
comparison table, FAQ accordion, and a CTA that opens a demo booking modal.
Deploy to a shareable Vercel URL."

How It Works

1. Architect

The agent analyzes your requirements, proposes a technical approach — stack, data model, component structure, API design — and presents it for approval before writing any code.

"Before building, lay out the technical architecture for this application:
recommended stack, database schema, key components, and deployment approach."

2. Build

The agent creates the project in a dedicated Daytona sandbox:

• Scaffolds the application
• Creates folder structure and config
• Implements components, API routes, and business logic
• Installs dependencies
• Commits each change to a feature branch

3. Live Preview

As the agent builds, a live browser preview opens in the Browser workbench panel. You see the actual application rendering — not a mockup — updating in real time as files change.

"Show me the current state of the app in the preview"

4. Iterate Conversationally

Describe changes directly — the agent updates the code and the preview refreshes:

"Make the sidebar collapsible"
"Add a loading skeleton to the data table"
"Change the primary button color to match our brand — #1a1aff"
"Add pagination to the customer list — 20 rows per page"

5. Test

"Write integration tests for the authentication flow and the
Stripe webhook handler. Run them and show me the results."

The agent writes tests, executes them in the sandbox, and reports results. If tests fail, the agent fixes the code until they pass.

6. Deploy

"Deploy to Vercel. Push to the main branch and give me the production URL."

The agent pushes the final code to GitHub, triggers a Vercel deployment, and returns the live URL.

Tech Stack Support

Common Use Cases

Internal Tools

Operations and finance teams get custom dashboards, approval workflows, and data entry forms in hours — without waiting in the engineering backlog.

"Build an internal tool where our ops team can view and approve
expense reports submitted via Airtable. Requires login. Deploy to Vercel."

Prototypes

Founders and PMs demo working products to investors and customers before engineering has written a line of code.

"Build a working demo of our AI writing assistant. GPT-4 integration,
live streaming responses, dark theme, mobile-responsive."

API Integrations

Connect two SaaS tools that don't talk to each other — the agent writes and deploys the integration.

"Build a webhook receiver that listens for new Stripe subscriptions,
looks up the customer in HubSpot, and creates an onboarding deal in the pipeline."

Data Applications

Analysts get a working web app to explore and visualize datasets — instead of static spreadsheets.

"Build a data explorer for our Postgres database. Let users query any table,
filter by date range, and export results as CSV. Read-only access."

Microservices

Build lightweight backend services without DevOps overhead.

"Build a PDF generation API — accepts HTML input, returns a rendered PDF.
Deploy to Railway, return the endpoint URL."

The Full Stack in One Workflow

"Build a complete customer feedback portal:
1. Public-facing form: name, email, feedback, rating
2. Stores submissions in Convex
3. Admin dashboard: view all feedback, filter by rating, mark as reviewed
4. Auto-email notification to support@company.com on new submission
5. Deploy to Vercel, share the live URL"

The agent handles every layer — frontend, database, auth, email, deployment — in a single connected workflow.

Tips

Be specific about the stack upfront. Specifying your preferred technologies prevents the agent from guessing:

• ✅ "Next.js 15, Tailwind, Convex, deploy to Vercel"
• ❌ "Build me a web app"

Use the preview to guide iteration. Watch the live preview and describe what to change as you see it — faster than writing detailed specs.

Ask for architecture before building complex apps. For anything beyond a simple CRUD app: "Before building, propose the architecture and get my approval".

Let the agent write the tests. After the build is done: "Write tests for the critical paths and run them".

Deploy early, deploy often. Even for prototypes: "Deploy the current state to a preview URL" so stakeholders can see real progress.

Common Questions

Does the code live in my GitHub? Yes. The agent commits to a GitHub repo. You own the code completely.

Can I open the project in my local IDE? Yes. Clone the repo the agent created and continue in VS Code, Cursor, or any editor.

What if I already have an existing codebase? Connect your GitHub repo in the chat input. The agent clones it and works in that context.

Can the agent deploy to my existing infrastructure? Yes, with access credentials: "Deploy to our AWS EC2 instance at this IP" or "Push to our existing Railway project".

How are secrets and environment variables handled? The agent creates a .env.example with all required variables and guides you through setting them in your deployment environment.