Frequently asked questions
Short answers about accounts, keys, cost, runs, quality and platforms.
Getting started#
Do I need an account?
No. There's no sign-up and no Rdn Labs account. Install the app, paste a provider API key into Settings, and start a run. See Quick Start.
Which providers are supported?
Eight: Anthropic, OpenAI, Google Gemini, DeepSeek, xAI, Meta, Kimi and Qwen. One key is enough to start. See the Provider reference for where to get each key.
Does my ChatGPT, Claude or X Premium subscription work?
No. Chat subscriptions don't include API access. You need an API key from the provider's developer platform, with billing or credit set up. See Add your first API key.
What phones does it run on?
Android 8.0 or later, and iPhone on iOS 17 or later. On phones the app is portrait-only.
Does it work offline?
You can browse your runs, files and reports offline. Running a build needs an internet connection, because every agent is a call to your AI provider.
Keys and privacy#
Where do my API keys go?
They're encrypted on your device: on Android with an AES-256-GCM key from the hardware-backed Android Keystore, on iPhone as device-only Keychain items. A key is only ever sent to its own provider's API. See Privacy and security.
Can Rdn Labs see my requirements or code?
No. Your requirements and generated code go directly from your phone to the AI provider you chose. Rdn Labs is never in the path. The only thing the app downloads from Rdn Labs is the public model and price list, and that request sends nothing about you.
Is my data backed up to the cloud?
No. On Android, cloud backup is disabled for the app. On iPhone, keys are device-only and the run database is excluded from iCloud backup. Export any run you want to keep.
Cost#
What does it cost to use?
You pay your AI provider directly, at its per-token prices. Rdn Labs takes no cut. A small project typically costs from a few cents to a couple of dollars, depending mostly on the models you choose. See Control what a run costs.
Can I cap what a run spends?
Yes. Set Max cost under Configuration → Orchestration. The swarm checks in before any iteration that could go over it. It's a predictive check, not a hard guarantee, so for a firm ceiling also set a spending limit in your provider's billing console.
Why doesn't the app's cost match my bill exactly?
The app estimates cost from published prices and the token counts the provider reports. Account discounts, credits, taxes and price changes can differ. Your provider's billing console is the authority. See troubleshooting.
Runs and quality#
Can I use more than one provider in a run?
Yes. Each of the nine roles can use a different provider and model. For example, the Project Manager can use one provider, the developers another, and QA a third.
How long does a run take?
It depends on the project, the models and the team size. A small project usually takes several minutes per iteration. Reasoning models and larger projects take longer.
What happens when QA finds bugs?
At the end of each iteration the swarm checks in with you. Tap Reject or Continue with a note, and the developers fix the files with open issues in the next iteration. With Autorun on, it keeps iterating by itself until the score reaches your threshold. See Iterations and check-ins.
Does a run stop when it reaches the score threshold?
No. The threshold decides when a run counts as Successful, and when Autorun starts checking in again. A run ends when you Accept at a check-in, when it reaches Max iterations, or when you stop at a Max cost check-in.
What is the Quality Scorecard?
Five dimensions: Correctness, Completeness, Integrity, Quality and Accessibility. Each is pulled down by its open issues. The overall Score is their weighted geometric mean, so one collapsed dimension sinks the whole Score. A run is Successful when the Score reaches your threshold with zero open critical issues. See The Quality Scorecard.
What kinds of apps can it build?
Single-page web apps in HTML, CSS and JavaScript work best. They can be previewed, tested and visually checked right on your phone. Other kinds of code can be written and reviewed, but not run on the phone. See Write good requirements.
Can I change the app after the run finishes?
Yes. Tap Run next iteration on the finished run and describe the change. It continues from the code it already has. See Continue a finished run.
What if my phone sleeps mid-run?
On Android, the run keeps going in the background with an ongoing notification. On iPhone, keep the app in the foreground for long runs, because iOS suspends background apps after a short grace period. If a run is interrupted, Run next iteration picks it back up.
Output#
Can I see and export what was built?
Yes. Output shows every generated file, the tests and any screenshots, and App Preview runs the app on your phone. Export the whole run as a zip from Output or History. See Review output and preview the app.
Who owns the code it generates?
You do, as between you and Rdn Labs. Your provider's terms may also apply. Review and test generated code before relying on it. See the Terms of Use.