AI Writer That Publishes to Your Own Domain: Setup Guide
AI Writer That Publishes to Your Own Domain: Setup Guide
If you’ve been waiting for an AI content tool that publishes to your domain instead of a SaaS dashboard, you’re not alone. Most “AI writers” are subscription platforms that own the publishing layer — you get a dashboard, they get the domain authority. Quilligator is a self-hosted AI content engine that drafts, edits, illustrates, and publishes affiliate articles directly to your domain, giving you full ownership of your content and domain authority. Here’s how to get it running.

Why Self-Hosted Publishing Matters
The standard SaaS content tool workflow looks like this: you write or generate an article, it lives in their CMS, they host it on their CDN, they own the domain. You pay monthly. If you leave, your articles stay behind a paywall or disappear.
A self-hosted setup flips the ownership model. You deploy the engine to your own hosting account (typically Railway, which handles Docker containers), point your domain at it, and the articles publish directly to your site. The bytes live on your volume. If you decide to leave or switch tools next year, you take every article with you.
This matters for affiliate publishers especially. Domain authority compounds over time. If you’re building a niche site you plan to own for years, you want that authority accruing to your domain, not a platform’s.
Prerequisites: What You’ll Need
Before starting, familiarize yourself with these tools:
- Railway — a container hosting platform that simplifies deployment (think of it as a simplified Heroku alternative). You’ll deploy Quilligator here.
- Docker — containerization technology that packages software so it runs consistently anywhere. Railway handles Docker for you; you don’t need to run commands yourself.
- API keys — credentials that let Quilligator call external services like OpenAI and Anthropic. These are text strings you’ll paste into Railway’s dashboard.
- DNS/CNAME — a domain name system record that points your domain to Railway’s servers. Your domain registrar’s control panel will have a DNS settings section.
If you’re unfamiliar with any of these, the setup guide walks you through each step. No command-line experience required.
The Core Architecture: What You’re Actually Running
Before setup, it helps to know what’s happening under the hood.
Quilligator ships as a single Docker image. When you deploy it to Railway, you’re spinning up a container that runs:
- The writer service — takes a topic, researches it, and drafts an article using a cost-optimized LLM (usually Claude Haiku for speed).
- The editor pass — a second LLM re-reads the draft, flags AI tells, hedging filler, and unsupported claims, then routes the article to either publication or human review based on quality gates.
- The illustrator — tries to find a relevant stock photo from Unsplash; if nothing fits, it can generate an AI image instead.
- The publisher — renders the article as static HTML, adds internal links, product cards, and a table of contents, then deploys it to your domain.
- The spend ledger — tracks API costs per site and per article, enforces budget caps so one runaway niche doesn’t drain another.
All of this runs on a schedule (typically one to three articles per day, configurable). The operator never touches the command line after deployment — everything is controlled from a web dashboard.
The key difference from SaaS: this is your infrastructure. Railway is just the host. You own the API keys, the article database, the domain DNS. Quilligator is licensed software you buy once and run forever.
Step 1: Buy the License and Prepare Your Environment
Head to the Quilligator Gumroad listing and purchase a license. It’s a one-time purchase with a fourteen-day refund window. Current pricing is available on the Gumroad page. You’ll receive a license key immediately.
While that’s processing, set up the infrastructure you’ll need:
Railway account: Sign up at Railway.sh. Railway is a container hosting platform that handles Docker deployments with minimal configuration. You’ll need a payment method on file (they charge per CPU-hour and GB-month of storage, but costs are low for a single Quilligator instance — typically /month).
Domain: Point a domain you own toward Railway. You can use a subdomain (e.g., content.mysite.com) or a full domain dedicated to your niche site. The DNS setup is straightforward — Railway gives you a CNAME target after deployment.
API keys: You’ll need: - An OpenAI API key (for image generation and some model calls). Create one at platform.openai.com. - An Anthropic API key (for the Claude models that power the writer and editor). Create one at console.anthropic.com. - Optionally, an Unsplash API key (for stock photo searches). Free tier is fine.
Keep these keys safe — don’t commit them to version control. Railway’s environment variable system will handle them securely.
Step 2: Deploy to Railway
Railway’s deployment process is designed to be fast. Here’s the flow:
- Log into Railway and create a new project.
- Connect your GitHub account (or upload the Docker image directly if you prefer).
- Paste the license key into the Railway environment variables section. Railway will prompt you for
QUILLIGATOR_LICENSE. - Add your API keys as environment variables:
OPENAI_API_KEY,ANTHROPIC_API_KEY,UNSPLASH_API_KEY. - Configure your domain. Railway gives you a temporary URL; update your DNS CNAME to point your domain at Railway’s endpoint.
- Deploy. Railway pulls the image, spins up the container, and your instance is live within a few minutes.
The entire process takes roughly fifteen minutes of active setup; DNS propagation takes 24–48 hours before your domain is live. During propagation, you can still access Quilligator via Railway’s temporary URL to configure your first niche.
Step 3: Configure Your First Niche Site
Once Quilligator is running, you’ll land in a dashboard. This is where you define what the engine writes about.
Create a site by giving it:
- A name (e.g., “Coffee Gear Reviews”)
- A slug (the URL path where articles will live, e.g., /coffee-reviews/)
- A budget cap (how much you’re willing to spend on API calls per month for this site)
Write a brand brief. This is the key differentiator. A brand brief is a short document (a few hundred words) that describes your niche, your target reader, tone preferences, and any products or claims you want to emphasize or avoid. Quilligator shows this brief to the writer on every article. Unlike SaaS tools that ship generic prompts, your brief ensures the engine writes for your audience, not a generic one.
Example brief excerpt:
“This site reviews mechanical keyboards for software developers. Our reader is technical, skeptical of marketing, and values build quality over RGB lighting. Avoid hype language like ‘gaming’ and ‘pro-grade.’ Always cite keyboard switch specs from the manufacturer. Recommend mid-range boards; we don’t review budget junk or luxury boutique models.”
Set your content clusters. A cluster is a pillar topic — “Best Mechanical Keyboards,” “Keyboard Stabilizers,” “Keycap Materials.” The engine uses these to generate keyword-relevant articles within your niche. You can add or rotate clusters anytime.
Link your affiliate programs. Quilligator has built-in product card rendering. When you add an Amazon affiliate ID, the engine can place relevant product recommendations in articles with live pricing and affiliate links. You can also add Awin, Impact, or custom affiliate networks.
Step 4: Set the Daily Publish Schedule
Configure how often the engine runs. Most operators start with one article per day. You can dial it up to three or back it down to three per week depending on your budget and how fast you want to build content.
Set a preferred publish time (e.g., 6 AM UTC). The engine will draft, edit, illustrate, and publish on that schedule automatically.
The spend ledger will warn you if a day’s API costs are trending high — if you’re burning through your budget too fast, the engine throttles itself to preserve runway.
Step 5: Monitor Quality and Iterate
The editor pass catches a meaningful minority of drafts that don’t meet the quality gate. These articles are held for human review in the dashboard instead of going live automatically. You can:
- Approve and publish the article as-is
- Request revisions (the engine will re-draft the weak sections)
- Reject the article and let the engine try a different topic
Expect 10–20% of drafts to be held for review in the first month as the engine learns your brand voice. This rate typically drops as you refine your brand brief. This human-in-the-loop mechanism is what separates Quilligator from “set and forget” tools. You retain editorial control. Over time, you’ll notice patterns — maybe the engine struggles with certain clusters, or overshoots on technical depth for your audience. Refine your brand brief, and the next batch of articles will reflect those changes.
Self-Hosted vs. SaaS Content Tools
If you’re weighing Quilligator against Jasper, Copy.ai, or Writesonic, here are the key differences:
Jasper (see pricing) has a more mature template library and a polished WYSIWYG editor, as of June 2026. If you want to write short-form content (ad copy, email subject lines, social posts) and don’t care about domain ownership, Jasper is solid. But Jasper publishes to their platform; you’re paying per month forever.
Writesonic (see pricing) has a lower entry tier if you publish infrequently (one article a month), as of June 2026. Quilligator’s economics favor operators publishing at least a few articles weekly.
Copy.ai is strong for bulk short-form generation. It’s not built for the affiliate-article publishing pipeline Quilligator specializes in.
WordPress + an AI plugin is a reasonable alternative if you already run WordPress. Quilligator’s value is that it handles research → draft → edit → illustrate → publish in one process without the WordPress operational overhead.
The core advantage of self-hosted: you own the domain, the articles, and the infrastructure. You’re not renting a dashboard. If you stay for three years, the domain authority is yours; the articles don’t evaporate if the company pivots or shuts down.
Handling API Costs and Budget Management
One anxiety new operators have: will an AI content engine drain my API budget?
Quilligator’s spend ledger is designed to prevent that. Each site has its own budget cap. When the engine approaches the limit, it throttles — it might skip a day or drop to shorter articles instead of stopping completely. This is different from SaaS tools, which charge a flat monthly fee regardless of usage.
In practice, generating one article per day per article based on Claude Haiku + OpenAI image generation at current API rates (June 2026). The spend ledger makes that visible and predictable.
If you’re worried about runaway costs, start with a conservative budget cap — say, enough for ten articles per month — and increase it once you’ve seen a month of real spending.
Customizing the Output: Headers, Footers, and Styling
By default, Quilligator articles include a byline, publish date, and table of contents. The dashboard lets you customize the output:
- Site header and footer — add your logo, navigation links, or a disclaimer.
- CSS overrides — hide the byline or publish date, tweak colors, fonts, and spacing to match your brand.
- Product card styling — control how affiliate product recommendations appear (card size, button color, etc.).
Example: if you want to remove the byline and publish date, you can add a CSS rule to hide them. If you want a custom color scheme, you can override the default stylesheet. Most operators keep styling minimal and let the engine’s defaults handle layout. The focus is on content quality, not design polish.
Troubleshooting Common Deployment Issues
“Railway says the container failed to start.” Check that your API keys are correctly pasted into the environment variables. A missing or malformed key will cause the service to crash on startup.
“Articles aren’t publishing.” Verify that your domain’s DNS CNAME is pointing to Railway’s endpoint. Use nslookup or a DNS checker to confirm. Also check that your brand brief is saved and your clusters are configured.
“The editor keeps rejecting drafts.” Expect 10–20% of drafts to be held for review in the first month. If the rate is higher, review rejected articles and identify the pattern (too much hedging? unsupported claims? wrong tone?). Refine your brand brief, and the next batch of articles will reflect those changes.
“API costs are higher than expected.” Check the spend ledger dashboard — it shows per-article costs. If costs spike, it’s usually because the writer is generating longer articles or the illustrator is falling back to AI image generation frequently. Adjust your brief to target shorter articles, or increase your Unsplash API quota.
Next Steps: Scaling to Multiple Niches
Once you have one niche running smoothly, Quilligator’s multi-site architecture lets you run two, three, or more niches on the same Railway deployment. Each site has its own:
- Domain (or subdomain)
- Budget ledger
- Brand brief
- Clusters and affiliate programs
You don’t pay per site — you pay once for the license, then host multiple sites on one Railway instance. This is a meaningful advantage over SaaS tools that charge per workspace or per site.
FAQ
Q: Does Quilligator support non-English languages?
A: The engine can draft articles in any language supported by Claude and GPT-4. Set your brand brief in your target language, and the writer will follow that voice. Stock photo search and image generation work in any language. Most operators use English; non-English support is available but less tested in production.
Q: Can I use my own LLM instead of Claude or OpenAI?
A: Not currently. Quilligator is optimized for Claude Haiku (writer) and GPT-4 (editor). You can’t swap in Llama, Mistral, or other open-source models without forking the codebase. This is a limitation if you want to avoid proprietary APIs.
Q: Can I edit articles after they’re published?
A: Yes. The dashboard has a simple editor for published articles. You can update text, swap the hero image, or adjust product recommendations. Changes are live immediately.
Q: What if I want to pause the engine for a month?
A: Set the daily publish schedule to zero articles per day. The engine will stop drafting but keep the dashboard and your article library intact. You can resume anytime.
Q: What happens if Quilligator shuts down or I decide to leave?
A: Your articles are static HTML files on your Railway volume. You can download them, export the metadata, and republish anywhere — WordPress, a static site generator, another host. You own the bytes.
Q: Do I need to know Docker or command-line tools?
A: No. Railway handles the Docker complexity. You just need to paste environment variables and click “Deploy.” The dashboard is a web interface.
Q: How long before articles start ranking?
A: SEO takes time. Most operators see meaningful traffic after two to three months of consistent publishing. Quilligator speeds up the writing labor, but Google’s ranking algorithm is still Google’s. Quality, topical depth, and backlinks matter.
Summary
Self-hosted AI publishing is no longer a technical novelty — it’s a practical alternative to SaaS content tools, especially for affiliate publishers who want to own their domain and data. Quilligator makes the setup straightforward: deploy in fifteen minutes, configure your niche, and let the engine handle the daily drafting and publishing work.
The tradeoff is that you’re responsible for hosting and monitoring. But the upside is real: no per-month SaaS fees, no platform lock-in, and domain authority that compounds on your site, not a vendor’s.
Ready to get started? Deploy Quilligator on Railway in fifteen minutes at https://quilligator.com.