Stop Using Firebase: The Definitive Guide to Deploying Google AI Studio Apps via Cloud Run

The architecture of modern AI application development has shifted. While Google AI Studio provides a rapid prototyping environment for building intelligent task flows and dashboards, moving these applications to a production environment remains a significant bottleneck for developers.

Common misconceptions suggest using Firebase Hosting or GitHub Pages. These methods invariably fail because AI Studio applications typically require a dynamic backend runtime—often Python or Streamlit-based—rather than a static site environment. The infrastructure simply doesn’t match the requirement.

The robust solution lies in Google Cloud Run. This serverless platform handles the containerization and serving of the application, allowing for a seamless transition from a prompt-based build to a live, scalable web application connected to a custom domain.

This technical manual documents the precise workflow for deploying an application from Google AI Studio, configuring the Google Cloud environment, and mapping a custom domain with full DNS propagation.

Phase 1: Initiating the Cloud Run Container

The deployment process begins directly within the AI Studio interface. Unlike traditional development lifecycles involving local CLI commands and Dockerfiles, AI Studio abstracts the containerization process into a single click-stream.

Step 1: The Deployment Trigger
Locate the Rocket Icon in the top navigation bar of the AI Studio interface. This initiates the deployment sequence.

• Select “Deploy app on Google Cloud”.
• The system will prompt for a project selection.

Step 2: Project Initialization
If a Google Cloud Project does not typically exist, one must be generated immediately.

1. Click “Create a new project”.
2. Assign a nomenclature, e.g., Your-AI-Taskflow.
3. Click Create.
4. Note: Project creation may take 30-60 seconds.

Step 3: Billing Authorization
Google Cloud Run is a paid service (though often within free tier limits for small apps), requiring an active billing account.

1. Upon project creation, the interface will likely prompt: “You must enable billing for this project to proceed.”
2. Click “Set up billing”. This redirects to the Google Cloud Console.
3. If a billing account exists, select “Link a billing account” and choose the active profile.
4. If no account exists, select “Manage billing accounts” to add a payment method.

Critical Troubleshooting: The “Refresh” Protocol
The integration between AI Studio and Google Cloud Console can be brittle. It is common for the AI Studio interface to fail to recognize the newly linked billing account immediately.

• The Fix: If the deployment modal hangs or throws an error after linking billing, perform a hard refresh of the browser tab running AI Studio.
• Re-click the Rocket Icon.
• Re-select the project.
• The system should now verify the project status.

Step 4: Executing Deployment
Once verified, the “Deploy app” button will become active. Click it.

• Observation: The console will display “Deploying to Cloud Run.” This process compiles the code, builds the container, and pushes it to the Google Cloud registry.
• Error Handling: If a “Deployment failed” error appears (common on first attempts), simply click “Redeploy app”. Persistence is often required here.

Upon success, a modal will display the App URL (e.g., https://your-ai-taskflow-601.us-west1.run.app). This is the live, public-facing endpoint.

Phase 2: Domain Acquisition and Management

While the run.app URL is functional, it lacks professional authority. The next phase involves acquiring a top-level domain (TLD) and preparing it for mapping.

The Registrar Choice
For this workflow, Spaceship is utilized due to its rapid DNS propagation, though the principles apply to Namecheap, GoDaddy, or Cloudflare.

1. Search for the desired domain (e.g., youraitaskflow.online).
2. Complete the purchase transaction.
3. Navigate to the Domain Manager or Launchpad.
4. Locate “Advanced DNS” or “Nameservers & DNS”. This is where the records will be modified.

Phase 3: The Cloud Run Handshake

The connection between the Google Cloud container and the registrar requires a specific mapping configuration within the Google Cloud Console.

Step 1: Accessing Cloud Run Networking

1. In the AI Studio deployment success modal, click the link labeled “Open in Google Cloud”.
2. This redirects to the specific Cloud Run service dashboard.
3. Locate the sub-navigation menu (Metrics, Logs, YAML) and click “Networking”.

Step 2: Initiating Domain Mapping

1. Under the “Integrations” or “Networking” tab, look for the “Custom domains” section.
2. Click “Manage”.
3. Select “Add Mapping”.

Step 3: Service Selection

• Service to map: Ensure the correct application (e.g., your-ai-taskflow) is selected.
• Domain selection: Click the dropdown. If the domain is not listed, select “Verify a new domain”.
• Base domain: Enter the naked domain (e.g., youraitaskflow.online)—do not include https:// or www.

Phase 4: DNS Verification (Google Search Console)

Google requires proof of ownership before allowing traffic routing. This is handled via Google Search Console using a TXT record.

1. Click “Verify in Search Console”. This opens a new window.
2. Select “Domain” property type.
3. Google will generate a specific TXT record string (e.g., google-site-verification=…).
4. Copy this string.

Updating Registrar Records:

1. Return to the Domain Registrar (Spaceship) Advanced DNS settings.
2. Add New Record:
   • Type: TXT
   • Host: @ (represents the root domain).
   • Value: Paste the Google verification string.
   • TTL: Automatic or 30 min.
3. Save the record.

Return to the Google Search Console tab and click “Verify”.

• Note: While often instant, this can take minutes. If it fails, wait 60 seconds and retry.
• Once ownership is verified (“Ownership auto verified”), close the Search Console tab.

Phase 5: Configuring A and AAAA Records

With ownership confirmed, the final step is routing the traffic.

1. Return to the Google Cloud Console “Add Mapping” window.
2. Click “Refresh” next to the domain list. The new domain should now appear as a verified option.
3. Select it and click “Continue”.

Google will now generate a list of required DNS records to point the domain to the Cloud Run load balancer. This typically includes:

• 4 ‘A’ Records (IPv4 addresses)
• 4 ‘AAAA’ Records (IPv6 addresses)

The Configuration Protocol:
Return to the Registrar’s DNS settings and input these records meticulously.

• For ‘A’ Records:
  • Type: A
  • Host: @
  • Value: [IP Address 1 from Google]
  • Repeat for all 4 unique IPv4 addresses.
• For ‘AAAA’ Records:
  • Type: AAAA
  • Host: @
  • Value: [IPv6 Address 1 from Google]
  • Repeat for all 4 unique IPv6 addresses.
• Technical Note: Some registrars may flag “conflicting records.” Ignore this warning if you are adding distinct IPs for the same host (@). This is standard load balancing.

Once all records are saved in the registrar, return to Google Cloud Console and click “Done”.

Propagation Latency:
The status may show as “Certificate Provisioning” or a spinner. DNS propagation generally takes 15–45 minutes, though it can take up to 24 hours. Once the SSL certificate is automatically provisioned by Google, the app will be live at the custom URL.

Phase 6: Continuous Deployment & Iteration

A production application is rarely static. When updates are required—for instance, adding a new “Email Integration” tab or modifying the UI—the deployment pipeline must be re-triggered.

1. Modify: Make the necessary code or prompt changes within Google AI Studio.
2. Redeploy: Click the Rocket Icon again.
3. Execute: Select the existing project and click “Redeploy app”.

Cloud Run utilizes revision handling. It will spin up a new container revision and traffic will be migrated to the new version automatically. The custom domain mapping remains intact; no further DNS changes are required after the initial setup.