Shipmunk FAQ
Straight answers to the AI app builder questions we hear most: what Shipmunk builds, how it tests, who owns the code and how credits work. Closed beta
Shipmunk FAQ: the basics
These are the questions people ask most before they try Shipmunk. Each answer describes the product as it works today. If you want the full picture first, start with how a Shipmunk build works, then come back here.
Common AI app builder questions
What is Shipmunk?
Shipmunk is an AI app builder. You describe an app in a sentence, answer a few questions and approve a plan. Shipmunk then writes the code, runs checks against it, fixes what fails and gives you a preview of the finished app. It is in closed beta today. The how it works page walks through each step.
What kinds of apps can I build?
Home opens on "Build a phone app". Under "Other kinds of app" you can build an MVP, add a feature to an existing repository, make a data pipeline, a document processing pipeline, an AI assistant or chatbot, or a dashboard with reports. There is also a "Something else" option for ideas that do not fit a template.
How do phone apps work, and what is Expo Go?
A phone app is a native React Native app made with Expo. You open it in Expo Go, a free app that runs Expo projects. On a phone, the iteration page gives you a link that opens the app in Expo Go. On a computer, you get a QR code to scan, and you can stream a cloud Android device with the app open for a short session. See phone apps for more.
What is the Sandbox?
The Sandbox is where a build is tested before you see it. When an app needs an outside tool such as Stripe or Slack, Shipmunk can use a realistic stand-in, marked "Simulated" on the page. The builder writes checks, runs them in a private container and fixes what fails. Read more on the Sandbox page.
Is it real code, and can I download it?
Yes. Each build is real source code, packaged as a zip. A finished iteration has a "Download code" button. If a build did not pass its checks, the button says "Download code (not passing)" so you know what you are getting. When you add a feature to a connected GitHub repository, Shipmunk opens a pull request on a new branch instead.
Can I change the app after it is built?
Yes. You can ask for a change to a ready or failed build. Shipmunk starts from the last round of that build, adjusts the plan for your request and runs a normal build, with the same checks. The new iteration is linked to the one it came from. Each change is a build of its own and uses credits.
How does billing work?
Builds use credits. Credits are charged for the AI and the compute a build uses, and you see your balance in the app. Before a build starts, Shipmunk checks that your balance covers it. If credits run out mid-build, the build pauses instead of failing, and topping up starts it again. Calls made on your own model key are not charged credits.
Can I bring my own model keys?
Yes. A workspace can store its own OpenAI or Anthropic key. Your browser encrypts the key before it is sent, and Shipmunk checks it with the provider before using it. Builder runs on your own key are not charged credits for those model calls. Planning questions still run on our key.
How safe are my builds and keys?
Each build runs in its own container. The builder's network access is fenced, so it cannot reach private or internal addresses. Every build has a hard time limit, and previews are removed when they expire. Secrets are masked before build traces are written. Stored keys are encrypted in your browser. The security page has the details.
Who is Shipmunk for?
Two kinds of builders. Non-technical founders who can describe an idea but cannot easily tell whether an app works. And product and engineering teams who want to try a feature against realistic tools and users before committing code to their own codebase. If you can describe what you want in a sentence, you can start.
How do I get access?
Shipmunk is in closed beta, so access is limited for now. Start at app.shipmunk.ai and sign in. Some features shown on this site, such as persona testing, carry a Closed beta label because they are not generally available yet. We do not promise dates for when they will be.
How is Shipmunk different from Claude Code or Codex?
Claude Code and Codex are great at writing code. Shipmunk answers a different question: does the app work when someone uses it? A diff tells you what changed. A Sandbox run shows what the app actually did. Shipmunk builds the app, then checks it before you see it.
Does Shipmunk launch my app to production?
No. The Sandbox is the rehearsal. Launching still needs your live accounts, any provider approvals and your own production checks. Shipmunk helps you reach that stage with fewer unknowns. You can download the code and take it wherever you choose to deploy it, or receive a feature as a pull request.
Still have a question?
For the testing side, read about the Sandbox and simulated tools. For how builds and keys are kept apart, read how Shipmunk keeps builds safe. You can check service health at status.shipmunk.ai.
Give the idea somewhere safe to fail.
Describe it in a sentence. Shipmunk builds it, then proves it in a sandbox.