[ NEURALARC_PRODUCT_—_TELLNOVA_STUDIO ]

Chat-to-App Builder with Live Preview

2026Developer PlatformsIn-depth

What it is

Describe an app in chat, watch it build and run in the browser.

Screenshot of Chat-to-App Builder with Live Preview — Chat-to-App Builder with Live Preview — live product

[ THE_PROBLEM ]

Why this existed

Non-engineers describe what they need in documents that get misread, and the feedback loop between idea and something runnable is measured in weeks.

[ WHAT_WE_BUILT ]

What we built

A conversational build surface with a real code editor and a container that runs the generated project in-browser — no local toolchain, no deploy step to see it work. Generated projects persist in a library with preview pages and can be exported as a zip. Around it, a credit system with metered consumption, subscription plans with recurring billing and webhook reconciliation, a reference system that can ingest PDFs as design input, an admin console over projects, payments, feedback and settings, and a generated sitemap.

Conversational build surface with a real code editor
In-browser container running the generated project with no local toolchain
Project library with preview pages and zip export
Credit system with metered consumption
Subscription plans with recurring billing and webhook reconciliation
Reference system ingesting PDFs as design input
Admin console over projects, payments, feedback and settings
Generated sitemap

[ HOW_IT_IS_USED ]

How a company uses it

Internal teams and non-engineers produce working tools rather than requirement documents — and the credit system means the business model is in place from the first user.

Built with

ConvexReactWebContainerMonacoRazorpay

[ COMMON_QUESTIONS ]

Questions clients ask

Why run the generated app in the browser?

Because the alternative is a deploy step, and a deploy step between "describe it" and "see it" destroys the feedback loop. Running in-browser means the result is visible in seconds with no local environment to install.

Why build credits in from the start?

Because inference is a real per-use cost that scales with generations, not accounts. A generous free tier with no metering is how AI products discover their unit economics too late. Metering from day one makes the business model testable.

Can generated projects be exported?

Yes, as a zip. Anything that traps output inside the platform limits it to prototypes — being able to hand the code to a developer is what makes it useful for real work.

[ RELATED_WORK ]