Node 22 HTTP services can use the managed runtime when they deploy from GitHub, use npm start, listen on the supplied PORT and expose /health.
AI-ready hosting for supported static websites
Publish your AI-built website from GitHub without running a server.
Host Luma is AI website hosting for supported static projects built with tools such as Claude Code, Codex and Cursor. Keep the source in GitHub, then publish to a permanent preview URL with HTTPS, deployment history and build logs.
For static HTML and supported React, Vite, Astro and Next.js static-export projects. Read the practical guides for Claude Code, Codex and Cursor.
Quick answer
What is AI App Hosting?
AI App Hosting is a managed deployment workflow for supported AI-built static websites and supported Node web applications. Your source stays in GitHub; Host Luma runs the appropriate build or managed Node release path and gives you a permanent hostluma.app preview URL with HTTPS.
Available today: GitHub-connected projects, static website publishing, supported build-time Node workflows, managed Node 22 web applications, permanent preview URLs, HTTPS, custom domains, CDN delivery, deployment history, build/runtime logs and version redeployment from a recorded successful Git commit.
Managed Node requirements: a non-empty
Managed Node requirements: a non-empty
npm start script, the supplied PORT and GET /health. PHP, databases, Docker and arbitrary workers are not included. Version redeployment runs through the normal publishing path; it is not a snapshot, database or environment restore. See the roadmap for the current product position.A clear fit
For teams that want to publish a website, not administer a server.
| You are | Host Luma is useful when | Check this first |
|---|---|---|
| A founder | You have a landing page, MVP front end or product site ready to share. | The published project must produce static output. |
| An agency | You need a repeatable GitHub-based route for campaign and brochure websites. | Keep client approval and repository ownership clear. |
| A developer using AI assistance | You want code in GitHub and do not want routine manual uploads. | Confirm static HTML or a supported static build. |
| A team moving a static export | You are assessing a static project rather than assuming every hosted runtime will transfer unchanged. | Review build output, APIs and server-side dependencies first. |
Compatibility
Can I use Host Luma?
The tool you used to create a project is less important than what the project needs at publish time. Host Luma AI Hosting supports static output and a deliberately bounded managed Node web-application path; it is not a runtime for every generated application.
| I built my website with... | Fit today | Why |
|---|---|---|
| Static HTML | Supported | Static HTML, CSS, JavaScript and assets fit the current publishing model directly. |
| Vite | Supported | Supported frontend projects can build static output with the current npm-based build path. |
| Astro | Supported | Astro is supported when the project builds to static output. |
| Next.js static export | Supported | Next.js is suitable when it is configured to export static files, rather than requiring server-side rendering. |
| React | Usually supported | React projects are a fit when they produce static files, such as a supported Vite or static build. |
| Claude Code, Cursor, ChatGPT or Gemini | Usually supported | These tools can help create a project. The resulting project still needs to meet the static-output requirement; Host Luma does not claim a product integration with each tool. |
| Lovable, Bolt or v0 | Needs assessment | Exported projects may be static or may rely on services outside the current public hosting model. Check the build output and runtime requirements first. |
| WordPress | Not suitable for AI App Hosting | WordPress needs PHP and a database. Use Managed WordPress Hosting instead. |
A quick rule: a static site is a good candidate when it can be built into browser-ready files. A supported Node web application is a good candidate when it can run as
npm start, listens on the supplied PORT and passes GET /health. PHP, databases, queues, Docker and arbitrary workers are outside this product.Before you start
You only need a few things to begin.
- A GitHub accountGitHub is the connected project route available today.
- A supported projectStart with static HTML or a project that can build to static output.
- A working buildRun the project’s normal build locally before publishing when it has one.
- An internet connectionThat is it. You do not need Linux, cPanel, SSH or FTP for the intended workflow.
An important boundary
Who AI Hosting is not for today.
Being clear here prevents the wrong purchase. Host Luma AI Hosting currently publishes supported static output; it is not a substitute for a general application platform.
These require a PHP application runtime and usually supporting services. They are outside the public AI Hosting workflow.
Use Host Luma Managed WordPress Hosting, which is the product designed for WordPress and its PHP/database requirements.
Containers, queues and background workers need a managed runtime designed for long-lived application processes.
If a request must run code on the server, choose a platform that explicitly supports that runtime and operational model.
Database-backed applications need an appropriate database service and application environment, not static-site publishing alone.
Not sure which category you are in? Check whether your project can generate a static output directory without relying on a server process. If not, contact Host Luma support before you buy; we will not position the current AI Hosting product as a fit for an unsupported runtime.
Why developers choose Host Luma
A defined publishing workflow with useful boundaries.
Keep source code in a repository and publish through the connected workflow instead of making manual file uploads the normal release process.
Host Luma serves static HTML, CSS, JavaScript and assets, and runs supported Node 22 web applications through its managed lifecycle. The two paths are selected from the project configuration, not inferred from a framework label alone.
The Portal provides deployment history, build information and readable logs, so a failed build does not become a guessing exercise.
Each AI Hosting project has a permanent hostluma.app preview URL with HTTPS for review and early sharing.
Do you need cPanel or Linux? No. The customer workflow is the Host Luma Portal and a connected project. Host Luma is a publishing interface, not a server console.
Deployment lifecycle
What happens after you click Deploy?
Deploy should not mean that something invisible happened. The practical developer loop is simple: build, commit, push, publish, preview and iterate. Each stage has a clear job.
Build
Create the site in your preferred editor or AI-assisted tool, then make sure it produces static files. For a frontend project, run its normal build locally first.
Commit
Record the change in GitHub. This keeps the source of the published version in a clear, reviewable place.
Push
Push the change to the connected GitHub project. GitHub is the connected provider available today.
Publish
Host Luma uses the appropriate static publishing path: direct static files or a supported npm-based build followed by static output.
Preview
Review the result on the permanent HTTPS preview URL and check deployment status or build logs in the Portal when needed.
Iterate
Correct the project in GitHub and publish the next change through the same route. With automatic publishing configured, future updates use the established workflow.
Supported now
Choose a project that matches the current hosting model.
Static websites
AvailableHTML, CSS, JavaScript and assets served as static output.
- Marketing websites
- Documentation sites
- Campaign pages
- Static product front ends
Build-time Node
AvailableStatic builds for supported React, Vite, Astro and Next.js static-export projects.
- Vite sites
- Astro sites
- React static builds
- Next.js static export
Managed Node web applications
AvailableSupported Node 22 HTTP applications deployed from GitHub, with npm start, the supplied PORT, /health, managed logs and restart controls. PHP, databases, Docker and arbitrary workers are not included.
Practical check: choose the static path when the build produces browser-ready files. Choose the managed Node path only for a supported HTTP application that meets the documented start,
PORT and health requirements. A polished interface does not make unsupported PHP, database, Docker or worker requirements available.Useful examples
Where the current workflow fits well.
Publish a static menu, location and booking-enquiry site from a connected repository. Any live booking service should remain with its provider.
Publish a static product site, waitlist or interface prototype while assessing any backend separately.
Share a personal portfolio or case-study site with a repeatable repository-to-preview publishing path.
Keep a campaign or product site in GitHub, build it through the supported path and share a preview URL for approval.
Publish static guides or documentation generated with an AI-assisted workflow, with a visible build result.
Use a connected repository and consistent publishing process for static brochure or campaign work.
Share a static front-end prototype through the preview URL; do not treat it as a hosted backend service.
Publish the static interface only when its service requirements are handled elsewhere; the current public product is not a custom API runtime.
Publish a static launch page, product explanation or early-access site without making a temporary upload folder your release process.
Use a generated static site as a starting point, keep it in GitHub and review each published version on the preview URL.
Before you publish
Avoid the assumptions that cause avoidable deployment problems.
Check for server-side rendering, APIs, databases, queues or persistent processes before deciding a project belongs in this workflow.
Preview URLs with HTTPS are available immediately. A custom domain is a separate step — attach yours from the portal when you're ready, with CDN delivery included automatically.
Use the project’s normal build command before publishing. The Portal logs help with hosting-side visibility, but they do not replace ownership of project code.
Keep the source in GitHub and use the connected deployment route, so the path from change to published result is clearer.
Pricing
Start with one AI Hosting project.
AI Hosting includes a 30-day free period, then costs £8 per month for 1 AI website, £12 for 2, or £18 for 5 under one subscription. Every plan includes the GitHub-connected publishing route, automatic publishing for connected projects, a permanent preview URL with HTTPS, deployment history, build logs and UK-based support.
Each included AI website can use a custom domain and global CDN delivery. Add further included websites from the portal without creating another subscription. For a previous successful release, the Portal can queue a new deployment from that recorded Git commit; it is not a snapshot, database or environment restore. See the main pricing section for the current purchase route and the roadmap for product status.
A factual hosting promise
Managed infrastructure with a customer-visible publishing path.
Host Luma is a UK hosting company that runs its own OVH dedicated infrastructure, with CloudLinux resource limits hand-tuned over 8+ months of operating real customer sites and Uptime Kuma watching every service around the clock. For AI Hosting, the product focuses on the practical route from a supported GitHub project to a published static website: connected source, managed build and publishing steps, an HTTPS preview URL, plus deployment status and logs in the Portal.
Support covers the hosting-side workflow, publishing status, preview URLs, HTTPS and account access, over live chat, WhatsApp or email. Your project code, project content and third-party services remain your responsibility. Read our content standards for how availability and product claims are reviewed.
Page owner: Host Luma editorial team. Technical review: Host Luma hosting team. Last reviewed: 1 August 2026. Product availability is checked against the public site and roadmap; no customer counts, uptime figures, awards or unsupported partner claims are used on this page.
Revision history
Version 3, 1 August 2026: added a compatibility guide, a clear non-fit section, a practical getting-started checklist, an explicit developer workflow and fact-checked developer questions. Version 2, 1 August 2026: established the current-product positioning, deployment lifecycle, use cases and factual availability notes.
Frequently asked questions
Common questions developers ask.
What is AI App Hosting?
It is a managed publishing workflow for supported AI-built static websites. Keep the source in GitHub, publish the static output and share an HTTPS preview URL.
Can I build with Claude, Codex, Cursor, Lovable or Bolt?
Yes, if the resulting project matches the supported static hosting model. These tools can help you create the project; Host Luma does not claim a product integration with each of them.
Do I need Linux, cPanel or FTP?
No. The intended customer workflow is the Portal and a connected GitHub project, rather than server administration or manual uploads.
Can I host a dynamic Node, PHP or API application?
Supported Node 22 HTTP applications can use the managed runtime when they deploy from GitHub, use npm start, listen on the supplied PORT and expose /health. PHP, databases, Docker and arbitrary workers are not included.
Do I get a custom domain?
Every project gets a permanent HTTPS preview URL, and you can attach one custom domain per project from the portal with global CDN delivery included.
What happens when a deployment fails?
Use the Portal’s deployment status and build logs to understand the publishing result, correct the project in GitHub and publish again through the same route.
Can I use npm?
Yes. The current supported frontend build path uses npm: it installs dependencies with npm and runs the project’s build script for supported projects.
Can I use pnpm or yarn?
They are not currently presented as supported build-package-manager contracts. The public build path uses npm, so use an npm-compatible project or ask support to assess the project before you rely on another package manager.
Can I use a private GitHub repository?
Private repositories can be configured through a per-application deploy-key flow. That access must be set up and verified for the application; do not assume a private repository will publish without the required GitHub access configuration.
Can I deploy from a GitHub organisation?
Organisation-owned repository access needs assessment and authorisation for the connected GitHub setup. It is not presented here as an unconditional self-service promise; contact support before relying on that workflow.
Can I redeploy?
Correct or update the project in GitHub and publish the next change through the connected workflow. For a recorded successful Git commit, the Portal can also queue a previous-version redeployment through the normal publishing path; it is not a snapshot, database or environment restore.
Can I delete a deployment?
We do not currently present a customer-facing deployment-deletion control as part of the public AI Hosting workflow. Contact support if you need help with an application or published project.
Can I reconnect GitHub?
Do not assume a self-service reconnect control. If a repository connection needs to change, contact support so the application access and deployment configuration can be reviewed safely.