From Bolt.new to production: GitHub, server, and stable deployment with ArduMaker
Published on 2026-10-27 by frfrand
How to move a Bolt.new app to your own hosting: export to GitHub, identify Bolt Cloud backend and configure VPS, SSL and deployments.
What it really means to leave Bolt
Bolt.new can create and publish an app quickly, even with managed resources. That's useful for validating ideas. Migrating to your own infrastructure only makes sense if you need to control runtimes, integrate systems, or meet requirements your current provider doesn't cover. It's not mandatory to migrate to have a real app: there is also managed production. What matters is knowing how to recover code, data, and processes.
What is exported through GitHub
The Git integration provides a basis for preserving code and history. Check package.json, lockfile, frontend type, and endpoints used. If the app uses database, identity, files, or functions managed by Bolt Cloud, identify their dependency. An exported repository does not imply that database tables have magically appeared on the VPS.
Case: AI-generated furniture catalog
A business publishes its products from Bolt and receives quote requests. After exporting GitHub, the frontend compiles, but the form still sends information to a function from the previous environment. In staging you set up a controlled endpoint, validate fields, add an email service, and verify that a real test request is received. Only then do you link the commercial domain. The goal is a complete experience, not showing a website that opens with buttons that have no effect.
Concrete deployment plan
1. Sync the project with GitHub and clone the commit you want to publish. 2. Identify runtime, build, functions, secrets, DB, and storage. 3. Decide which pieces you keep in Bolt and which ones you move to ArduMaker. 4. Set up staging with the server configuration your stack needs. 5. Verify login, forms, emails, files, and errors. 6. Configure HTTPS, logs, backup, and rollback before announcing the change.
Differences: Bolt Cloud and a VPS
Bolt Cloud reduces infrastructure work; a VPS allows greater operational control, but you have to manage it. Neither replaces auditing the code. A landing page that only displays information does not need the same architecture as an ERP with hundreds of transactions.
Frequently asked questions
### Do I have to rewrite everything? Not necessarily. First check which code and services are portable. ### Can I continue creating with Bolt? Yes, as long as you document the sync and publish flow and avoid overwriting changes. ### Is npm run dev enough? No. Define the build and service suitable for production. ### Does the database move on its own with GitHub? No: data exports and secrets are independent processes.
Sources
Bolt: Introduction and Bolt Cloud · External deployment guide.
ArduMaker and the future migration service
Once the code has been audited, ArduMaker can serve as deployment infrastructure according to the stack: VPS, services, domain, certificates, and operational tools. This does not guarantee that the AI-generated application is already secure, nor that moving data is automatic. The page ardumaker.com/lovable is being prepared to present a migration and deployment service from AI builders to production environments; it is not yet published. For now, you can learn about the platform at ardumaker.com. The decision to migrate should be based on concrete requirements, not on a supposed obligation to abandon the current provider.
Updated: October 2026. The features, plans, and exports of these platforms may change. Check the current documentation and back up real data before intervening.
What the photograph shows
Server room ready to publish an application. Illustrative photograph generated to accompany this guide on migrating applications created with AI; it is not a capture of the platform nor does it certify a specific installation.
Continue with these practical guides from ArduMaker
- Migrate Bolt Cloud data to PostgreSQL: orders, users, and storage - Migrate Replit Agent to your VPS: processes, secrets, GitHub, and custom domain - Checklist: 30 verifications before publishing an app made with Lovable or AI
If you are evaluating your own application, compare the real backend, database, authorization, and deployment requirements before choosing the server.