About

How it works

From a prompt to a running full-stack app — the PrettiFlow request lifecycle and the difference between Plan and Build modes.

PrettiFlow turns a prompt into a running application through a multi-step agent loop. You don't manage any of it — but here's what happens under the hood.

The journey of a prompt

You send a prompt

Describe what you want from the PrettiFlow workspace. Optionally attach images, connect a GitHub repo, or use voice.

PrettiFlow understands intent

A fast model reads your prompt, figures out what you're asking for, and builds context from your workspace.

It plans the work

The agent drafts a plan — the files to create, the steps to take, and the order to do them in.

It builds in a sandbox

PrettiFlow spins up a secure cloud sandbox and writes, runs, and tests real code there. Multiple specialized agents can work in parallel.

Results stream back live

You watch progress in real time as files appear and the app comes together.

You export or deploy

Push the result to GitHub or deploy it live — the code is yours.

Plan mode vs. Build mode

PrettiFlow gives you two ways to work:

Plan mode

Suggestion-first. PrettiFlow proposes what it intends to do so you can review and steer before any code is written. Best for bigger or riskier changes.

Build mode

Direct. PrettiFlow implements your request straight away. Best when you know exactly what you want and want it fast.
Use Plan mode when scoping a new feature, and Build mode for quick, well-defined edits.
Copyright © 2026