Home › Services › GitHub Pages websites
GitHub Pages websites: fast static sites built for AI search
A GitHub Pages website is a complete website made of plain HTML files, hosted free on GitHub, and handed to its owner outright. Gerris designs and builds them for businesses that need a real website—one that loads fast, states their facts clearly, and reads cleanly to the AI answer engines—without a content management system to feed, patch, and defend. The product is the finished site: your domain, your repository, your files, with the claim layer for AI search built into the first page rather than bolted on later.
What is a GitHub Pages website?
GitHub Pages is GitHub’s hosting service for static websites. The site lives in a repository as finished pages: HTML, styles, and images, served directly to every visitor who asks. There is no database behind it, no server-side code, and no admin panel to secure. Your domain points at it, HTTPS is included, and the repository itself is the website in a form any developer can pick up and work on. Ownership is literal: whoever holds the repository holds the site.
That simplicity is the entire design. A conventional website is a machine that assembles pages on demand, and the machine needs fuel, updates, and guarding for as long as it runs. A static site skips the machine. The pages already exist, complete, and the host hands them over. For the brochure-shaped web—service firms, practices, portfolios, publications, and local businesses whose site exists to be found, read, and believed—the finished page is all the machinery the job requires.
Why do static websites load so fast?
A dynamic site builds each page at the moment of the request: database queries, a chain of plugins, a theme rendering the result. A static site does none of that work at request time, because the page was finished before anyone asked for it. A visitor asks for a page and gets the page. Speed here is structural, a property of what the site is, and it shows up everywhere speed is measured: Core Web Vitals scores, the share of your site a crawler can fetch in one visit, and the patience of a buyer on a phone.
The same structure removes the usual failure list. There are no plugins to fall out of date, no database to corrupt, no login page for attackers to knock on. The maintenance a static site asks for is the maintenance of its facts: services, prices, people, and places, kept current the way your Google Business Profile is kept current.
Why is a static site the right shape for AI search?
AI answer engines read the public web, and plain HTML is the easiest material they are ever handed: complete pages that any crawler can fetch end to end, with the facts stated in text rather than locked inside scripts or images. Gerris builds every site to be read that way. Chris calls the goal AI catnip: pages the answer engines find effortless to read, verify, and quote.
The claim layer goes in at the build, as part of the product:
- Entity schema: structured data that ties the business to its profiles, its people, and its services as one entity, with sameAs links the engines use to corroborate the record.
- An llms.txt file: a plain text statement of who the business is and where its key pages live, written for language models the way a sitemap is written for search crawlers.
- Question-led pages: headings in the form buyers ask their questions, with the answer stated directly under each one, in the buyer’s own words for the service and the place.
- FAQ content with matching schema: real questions answered on the visible page, in sentences an engine can lift without rewriting them.
- A welcoming robots.txt and a clean sitemap: the AI crawlers invited by name, and every page on the site accounted for in the index.
A business that launches with this layer in place starts its AI record as the author of it. The engines summarize every business they can find; a site built this way hands them the summary, in the business’s own facts, from the first day.
What does a Gerris build include?
The delivered product is the complete, launched website:
- Design and pages: the layout, the page set, and the copy structure—category stated plainly, services in buyer language, one canonical statement of what the business is.
- The claim layer: entity schema, FAQ schema, llms.txt and llms-full.txt, sitemap, and robots.txt, as described above.
- Domain and DNS: the domain connected, the records set, HTTPS live.
- Measurement in the owner’s name: Search Console, Bing Webmaster Tools, and analytics registered to you, so the evidence about your site belongs to you.
- Launch verification: the finished site checked the way the engines will meet it—crawlable, indexed, consistent with your Google Business Profile and listings.
This site is one of the builds. gerris.ai runs on GitHub Pages with its llms.txt and schema in the open, and everything claimed on this page can be inspected on it directly.
Can I update the website myself?
Yes. Self-service is part of the product, and it is why the handover is outright. The files are plain HTML in your repository. You can update the text in your own voice, with your own prompts and tools, whenever the facts change—a new service, a new hire, a new location—and the site republishes as fast as you can save the file. Any developer can work on it too; nothing is locked inside a builder that only Gerris can open, and no monthly fee keeps your own website available to you. For changes you would rather hand over, and for the ongoing AI visibility work the site was built to support, Gerris remains on call.
What does a GitHub Pages website cost?
The build is scoped and priced in writing before work begins, in conversation; it is not one of the published-price offers, like the AI Visibility Audit. The running costs are the short list: hosting on GitHub Pages is free, and the domain costs about $10 a year, paid by you directly to the registrar. There are no licence fees, no plugin subscriptions, and no hosting bill.
Which websites do not belong on GitHub Pages?
Online stores with large catalogues and checkout, web applications, and membership sites with logged-in areas need platforms built for those jobs. Chris says so on the opening call when a project points that way. A static site is the right answer for the site most businesses actually need: the one that states what they do, where they do it, and why a buyer should call—and that the answer engines can read as easily as the buyer can.
Frequently asked questions
Is GitHub Pages hosting really free?
Yes. GitHub serves static websites from a repository at no charge. Your running costs are the domain, about $10 a year paid to your registrar, and any changes you commission after launch.
Can AI answer engines read a GitHub Pages website?
Yes, plainly. The pages are finished HTML that any crawler can fetch end to end, and Gerris builds in entity schema, an llms.txt file, and question-led pages from the first day, so Google AI Overviews, ChatGPT, Perplexity, and Gemini read the same record your visitors do.
What happens when I want changes later?
You own the repository, so you can edit the files yourself, hand them to any developer, or ask Gerris to make the change. Nothing about the site depends on Gerris staying in the room.
Is a GitHub Pages website good for a local business?
Yes. The site states the services, the service area, and the facts the Google Business Profile repeats, and that one consistent record feeds the map results and the AI answers alike.
Read next: all Gerris services, the guide to Jekyll or plain HTML on GitHub Pages, and the guide to answer engine optimization.