If Claude Code, ChatGPT, OpenCode, Cursor, or another AI coding assistant has already built your website, you can publish it with Deplion without moving the project through a separate manual workflow.
The short answer: connect Deplion to your AI assistant, ask it to publish the current website, let it connect the existing contact form, and then open the public URL it returns. After publication, the same AI assistant can update the site, publish new versions, roll back an unwanted change, and read incoming leads.
The workflow is:
Build with AI → publish with AI → get a public URL → receive leads → keep updating with AI.
This guide shows exactly what you need to do yourself.
Before you start
You need:
- a website that is already built or close to finished;
- Claude Code, ChatGPT, OpenCode, Cursor, VS Code, Windsurf, or another MCP-compatible AI tool;
- a Deplion account;
- Deplion connected to your AI assistant.
For the fastest path, finish the website first and make sure the version you want to publish looks correct.
Deplion is intended for static websites and landing pages made from frontend files such as HTML, CSS, JavaScript, images, and other browser assets. If your project depends on its own backend runtime or database, see the limitations section below before publishing.
The fastest way: use one prompt
If Deplion is already connected to your AI assistant, start here:
Publish this website with Deplion. Keep the current design. Make the existing contact form collect leads. Check that the page works correctly on desktop and mobile, then give me the public URL.
When the AI finishes:
- Open the URL it gives you.
- Check the page on desktop and mobile.
- Submit the contact form once yourself.
- Ask the AI to show you the latest lead.
If those four checks pass, the website is ready to share.
Step 1: Connect Deplion to your AI assistant
Connect Deplion to the AI tool you already use.
Deplion works through MCP, so a compatible AI assistant can work with your Deplion projects directly from the same conversation or coding session where you are building the website.
You normally only need to connect it once.
After the connection is available, you should be able to ask the assistant to create and publish a Deplion project in plain language.
A simple first check is:
Can you use Deplion to publish this project?
If the assistant can see Deplion, continue to the next step.
Step 2: Finish the website before publishing
Build the website the same way you normally would.
For example:
Create a landing page for my accounting firm with a hero section, services, testimonials, FAQ, and a consultation form.
Then make the usual changes:
Make the hero shorter.
Improve the mobile layout.
Add three customer testimonials.
Change the main CTA to “Book a consultation”.
You do not need to learn a special publishing workflow while designing the page. Get the website into the state you want people to see first.
Step 3: Ask the AI to publish the website
When you are happy with the site, tell the assistant:
Publish this website with Deplion and give me the public URL.
The AI handles the Deplion publishing steps for you.
When it finishes, you receive a public URL that you can open in a browser and share with other people.
For the user, this replaces the repetitive workflow of manually packaging files, moving them into another service, uploading them again after every change, and keeping track of which version is currently public.
Step 4: Make the contact form actually collect leads
A contact form is only useful if the submission goes somewhere.
If your website already contains a contact form, tell the AI:
Make the existing contact form collect leads through Deplion. Keep its current design.
The important part for you is that the form can continue looking like part of your website while Deplion receives its submissions.
A typical lead form might contain:
- name;
- email;
- phone;
- company;
- service of interest;
- message or project description.
You do not need to manually build a separate form-processing application for a normal landing page.
Step 5: Test the website like a real visitor
A public URL is not the final check.
Open the website yourself and go through the same path a visitor would take.
Check:
- Does the page load correctly?
- Does it look right on mobile?
- Do the navigation and buttons work?
- Does the contact form submit successfully?
- Does the lead appear in Deplion?
For a quick test, submit something obvious:
Name: John Smith
Email: [email protected]
Message: Testing the contact form before launch.
Then ask your AI assistant:
Show me the latest lead from this website.
If it returns the test submission, you have verified the complete flow from visitor to lead.
Step 6: Read incoming leads through your AI assistant
Once people start using the form, you can ask the assistant for recent submissions.
Examples:
Show me the latest leads from this website.
Show me the last five inquiries.
Did anyone contact us today?
Summarize the newest inquiries and tell me what each person is asking for.
That closes the useful loop:
AI builds the page → visitor submits the form → the lead is available to the AI.
How to publish a Claude Code website with Deplion
If your project is already open in Claude Code and Deplion is connected, you do not need to export it to another workflow first.
Use a prompt such as:
Publish the current website with Deplion. Keep the existing design, make the contact form collect leads, verify that the site works correctly, and return the public URL.
After the first publication, you can continue working in the same Claude Code project and ask it to publish updates whenever you are ready.
How to publish a ChatGPT website with Deplion
If you created or edited the website in ChatGPT and Deplion is available in the conversation, ask:
Publish this website with Deplion. Preserve the current layout, connect the contact form to lead collection, and give me the public URL.
Once it is public, you can continue asking ChatGPT to change the page and publish the next version.
How to publish an OpenCode website with Deplion
If OpenCode is working with the project files and Deplion is connected, ask:
Publish the current project with Deplion. Keep the current design, make the contact form work, check the result, and give me the public URL.
The same pattern applies to later updates: make the change, review it, then ask OpenCode to publish the new version.
How do I update the website later?
You do not need to repeat the original setup every time the site changes.
For example:
Change the hero headline to “Launch your project faster.”
Review the result, then say:
Publish the updated version.
You can use the same workflow for pricing changes, new testimonials, image replacements, new sections, copy edits, or mobile improvements.
What if the AI changes the wrong thing?
AI-assisted editing is fast, but eventually an assistant will misunderstand a request or make a version you prefer less.
Typical examples include:
- the wrong section changes;
- a mobile layout gets worse;
- content disappears;
- the assistant rewrites more than you asked it to;
- the new version simply does not look as good as the previous one.
Deplion keeps published versions of the project.
If the latest publication is wrong, ask:
Roll back to the previous version.
This is more reliable than asking the AI to reconstruct the old page from conversation history.
Common problems and what to do
The AI cannot see Deplion
The Deplion connection is probably not available in the current AI client or session.
Check that Deplion has been connected to the MCP-compatible assistant you are using, then ask again:
Can you use Deplion in this session?
The AI published the site but the form does not work
Ask the assistant to verify the published form rather than redesigning it:
Check the published contact form and make sure it is connected to Deplion lead collection. Keep the existing visual design, fix the form if needed, republish the website, and then tell me when I can test it again.
After that, submit another test lead yourself.
The website looks different after publication
Ask the AI to compare the project with the published result and check that all required assets are included:
Check the published website against the current project. Fix missing styles, scripts, images, or other assets without redesigning the page, then publish the corrected version.
The latest AI edit made the site worse
Use version history instead of trying to manually undo multiple changes:
Roll back to the previous published version.
Then make the requested change again from the working version.
My project has a backend or database
Deplion is designed for static websites and landing pages. If the project depends on its own server-side application, database, authentication backend, or another runtime that must execute outside the browser, that part requires a different solution.
A frontend can still be suitable for Deplion, but do not assume a full-stack application will work unchanged.
Can I use my own domain?
Yes, Deplion supports publishing projects on a custom domain depending on the plan available to your account.
A practical workflow is to publish and test the site first, then connect the domain when you are ready to use it publicly.
Do I need GitHub to publish an AI-generated website?
Not for the Deplion workflow described in this guide.
If your AI assistant already has access to the project and Deplion is connected, the assistant can publish the website through Deplion directly.
You can still use GitHub for source control if it is useful to your development process, but it is not a required intermediate step for this publishing workflow.
Do I need to manually upload the files?
Not when the AI assistant already has access to the project and can use Deplion.
You can ask the assistant to publish the current website and let it handle the project files and assets required for publication.
Will my contact form work on a static website?
Yes, a Deplion-published landing page can collect form submissions as leads without you building a separate backend just for the form.
After publishing, always submit one test lead yourself before sharing the site publicly.
Can the AI show me the leads later?
Yes. You can ask a connected AI assistant for recent leads from the project.
For example:
Show me the latest leads from this website.
This is useful when the same assistant is already helping you maintain the page.
Can I publish another version after making changes?
Yes.
Make the change with your AI assistant, review it, and then ask:
Publish the updated version.
Deplion keeps published versions, which also gives you a rollback path if the new version is not what you wanted.
A practical example from start to finish
Imagine you are launching a small consulting business.
First you ask your AI assistant:
Create a landing page for an operations consultant. Include services, three case studies, an FAQ, and a consultation request form.
You refine the page until it looks right.
Then:
Publish this website with Deplion. Keep the current design, make the consultation form collect leads, check desktop and mobile, and give me the public URL.
You open the link and submit a test request.
Then:
Show me the latest lead.
The next day, you add another service:
Add “Process Automation” to the services section.
After reviewing it:
Publish the updated version.
If the new edit turns out worse:
Roll back to the previous version.
That is the full operating cycle without turning publication into a separate repetitive task.
One prompt worth saving
For most small landing pages, save this prompt:
Publish this website with Deplion.
Keep the existing visual design.
Make the current contact form collect leads.
Check that the website works correctly on desktop and mobile and that all required assets are included.
Publish the finished version and give me the public URL.
Then test the form once yourself.
Key takeaways
- If your AI has already built the website, you can ask the same assistant to publish it with Deplion.
- A public URL is only half the job; test the contact form and verify that a lead actually arrives.
- You can ask the AI to read new leads after publication.
- Later edits can be published from the same workflow.
- If an AI edit goes wrong, use the previous published version instead of trying to reconstruct it manually.
- Deplion is best suited to static websites and landing pages; projects that require their own backend runtime need a different architecture for that backend.
The goal is not just to end up with generated code.
The useful result is a website people can open, a form they can submit, and a project your AI assistant can continue maintaining after launch.