How to configure the knowledge base on a subdomain

The Semly knowledge base is a separate space with articles that help AI models better understand your brand, products, services, advantages, and answers to customer questions.

It can run under your own subdomain, for example:

knowledge.yourdomain.com

This way, the content is publicly available for users and LLM bots, while also being optimized for how AI models analyze websites.

You can see an example knowledge base here:

https://knowledge.semly.ai/

What is the Semly knowledge base

The Semly knowledge base is a public page with articles prepared based on an analysis of the brand’s visibility in AI.

Its goal is to organize information about the brand in a way that’s understandable for:

• users

• AI models

• LLM bots

• search engines

• systems that analyze content

• tools that use structured data

A knowledge base can include articles that answer customer questions, product explanations, guides, comparisons, FAQs, and content that supports specific prompts monitored in Semly.

A knowledge base isn’t just a place to publish articles. It’s an organized layer of knowledge about the brand, prepared for users and AI systems.

Why the knowledge base runs on a subdomain

The knowledge base can be launched on your subdomain, for example:

knowledge.yourdomain.com

This way, the content is tied to your brand and domain, but at the same time can work in a separate, technically optimized space.

This solution is useful when:

• you don’t have easy access to the CMS

• publishing on the blog requires a developer’s work

• you want to separate AI content from the main blog

• you want to publish articles generated in Semly faster

• you want to keep a consistent content structure

• you want to have a ready-made knowledge layer optimized for LLM bots

• you don’t want to interfere with the current template or website technology

A subdomain lets you quickly launch a knowledge base without rebuilding your existing website.

The knowledge base runs on Semly’s infrastructure

The knowledge base is hosted on Semly’s servers. This means you don’t have to build a publishing system yourself, implement a template, create the technical structure, or configure data for AI models from scratch.

Semly takes care of the technical layer of the knowledge base, including:

• content hosting

• article structure

• automatic sitemap

• llms.txt file

• structured data

• AI-friendly content formats

• public access to content

• optimization for LLM bots

• API

• public MCP for LLM bots

Thanks to this, the client can focus on content quality, correctness of information, and linking to their own products or services.

The knowledge base is set up so that it’s easy to read both for users and for AI systems analyzing the content.

What is generated automatically

Once the knowledge base is configured, Semly automatically prepares technical elements that help organize the content and make it available to AI systems.

Automatically generated include, among others:

• XML sitemap

• llms.txt file

• structured data

• article structure

• links to content

• content versions in AI-friendly formats

• JSON-LD data

• content in Markdown

• public article URLs

• structure available for LLM bots

Thanks to the automatic sitemap and the llms.txt file, AI models can more easily find and understand the most important content in the knowledge base.

Why llms.txt matters

The llms.txt file is a simple text file that helps AI systems understand what a given site is and where its most important resources are located.

In the Semly knowledge base, the llms.txt file can point to, among other things:

• a description of the knowledge base

• a list of the most important articles

• a link to the sitemap

• key resources for AI models

• brand information

• the language and context of the content

Thanks to this, LLM bots can more quickly identify that the knowledge base is a place with organized information about the brand.

The llms.txt file doesn’t replace good content, but it helps AI systems find and interpret the most important resources faster.

Structured data and AI-friendly formats

The Semly knowledge base is built with AI models and content-analyzing systems in mind.

Articles can be shared in formats such as:

• HTML

• Markdown

• JSON-LD

• structured data

• API

• public MCP

Structured data helps better describe the type of content, its topic, context, and relationships between elements.

Markdown is easy for AI systems to read and process.

JSON-LD lets you pass data in an organized form that’s useful for systems analyzing the structure of the content.

The Semly knowledge base connects content for the user with a technical structure that’s friendly to AI models.

Public API and MCP for LLM bots

The Semly knowledge base can also share content via open API and public MCP.

This way, AI systems and LLM bots can access structured data in a more technical and predictable way than by classic page browsing.

This can make it easier to:

• read the list of articles

• fetch content in a structured format

• analyze brand knowledge

• access data in Markdown or JSON-LD

• let AI systems use the knowledge base

• interpret relationships between articles and topics

Public APIs and MCP help share brand knowledge in a way that’s friendly to modern AI systems.

How to set up a knowledge base on a subdomain

To run a knowledge base on your own subdomain, you need to add the right DNS record.

Example subdomain:

knowledge.yourdomain.com

Basic steps:

1. Go to the knowledge base configuration in Semly

2. Enter the subdomain name, e.g. knowledge

3. Copy the DNS record value shown in the panel

4. Log in to the DNS panel of your domain

5. Add the required record for the subdomain

6. Save the changes

7. Go back to Semly

8. Click Verify and activate

Don’t change the main domain record if it serves your current website. Add a separate record only for the selected subdomain.

Example subdomain configuration

If your domain is:

yourdomain.com

you can create a subdomain:

knowledge.yourdomain.com

In the DNS panel, you add a record for the subdomain itself knowledge, not for the entire domain.

Example:

Type: A
Name: knowledge
Value: IP address provided in the Semly panel

After saving the changes, DNS may need some time to propagate. Usually it takes from a few minutes to a few hours, depending on the domain provider.

If you don’t have access to the DNS panel, pass the record details to a tech person or the company managing your domain.

What the buttons look like in the knowledge base

In the knowledge base you can set up buttons that help the user get to the most important places.

Most often, two buttons are used:

• a button leading to the knowledge base home page

• a button leading to the client’s landing page

Example:

• Knowledge base leads to the knowledge base home page

• Go to store leads to the store’s home page or a selected landing page

You can adjust the button texts to match your own communication style.

Example button texts:

• Go back to the knowledge base

• Go to store

• See offer

• Check products

• Contact us

• Visit page

• See services

One button should help the user move around the knowledge base, and the other should lead to the main page, store, offer, or another landing page.

How to add your own HTML

The knowledge base lets the customer adjust selected elements of the page by adding their own HTML code.

You can add HTML, among other places:

• in the article list

• in the footer

• above the content of all articles

This way you can add extra information, links, messages, navigation elements, or simple promo blocks.

Example uses:

• information about the brand

• link to the store

• link to the contact form

• additional CTA

• short offer description

• link to the main page

• information about help

• section with the most important categories

When adding your own HTML, make sure the code is correct and doesn’t make it harder for users or AI bots to read the content.

Internal linking to the landing page

Articles in the knowledge base should include links to the client’s most important pages.

Internal linking helps the user move from educational content to a specific product, service, category, or contact form.

In articles, it’s worth linking to:

• home page

• category pages

• product pages

• service pages

• contact form

• guides

• FAQ

• delivery policy

• returns policy

• comparison pages

Example:

The article answers the question:

It’s worth adding links to:

• category of beds with storage

• selected products

• guide to choosing a bed

• page with delivery information

Internal linking makes the knowledge base not just a collection of articles, but guides the user to the right products and services.

Uniqueness of articles and duplicate content

Articles published in the knowledge base should be unique.

You shouldn’t copy 1:1 content that’s already published on the blog, category page or other subpages. This can lead to a duplicate content problem, meaning the same content is repeated in many places.

Instead of copying existing content:

• prepare a new article for a specific prompt

• describe the topic from a different perspective

• add more purchase-related context

• answer specific customer questions

• link to existing pages instead of copying their content

• update articles based on recommendations

• avoid repeating the same paragraphs

Don’t copy content 1:1 from the blog or product page into the knowledge base. Articles should be unique and answer specific user questions.

The knowledge base is available to bots and users

The knowledge base is public. This means that both LLM bots and regular users can use it.

That’s why the content should be:

• useful

• understandable

• up to date

• factually correct

• unique

• well linked

• aligned with the offer

• written in natural language

• readable for humans

• organized for AI systems

Content in the knowledge base should be written for both users and AI models at the same time. The best articles are helpful for humans and technically easy to understand for AI systems.

Add a link to the knowledge base on your site

Once you launch the knowledge base, it’s a good idea to add a link to it from the client’s main page.

The most common places for the link:

• page footer

• menu

• help section

• blog

• knowledge center

• contact page

• FAQ page

• HTML sitemap

The simplest solution is to add a link in the footer, for example:

Baza wiedzy

leading to:

https://knowledge.yourdomain.com

This way, users and bots can find the knowledge base more easily.

Adding a link to the knowledge base from the main page boosts its visibility and helps bots find new content.

Add the knowledge base to the XML sitemap

Apart from the link in the footer, it’s worth adding the knowledge base to the main XML sitemap.

Thanks to this, systems analyzing the site can detect faster that the knowledge base is part of the brand’s ecosystem.

In the main sitemap you can add the address:

https://knowledge.yourdomain.com

or a reference to the knowledge base sitemap, if it’s available.

If you don’t have access to the XML sitemap, ask a technical person to add the knowledge base to the sitemap or to link to it from the main website.

How to check if the knowledge base works correctly

After configuration, check a few basic elements.

Checklist:

• the subdomain opens in the browser

• the knowledge base homepage works

• articles are publicly available

• buttons lead to the correct addresses

• links in articles work

• the sitemap is available

• the llms.txt file is available

• the content isn’t a copy of existing articles

• the knowledge base is linked from the main page

• the knowledge base has been added to the XML sitemap

• custom HTML doesn’t break the page layout

• articles include links to products or services

After launching the knowledge base, check it both technically and from the user’s perspective.

What to avoid when setting up your knowledge base

The most common mistakes are:

• no link to the knowledge base from the main page

• not adding the knowledge base to the XML sitemap

• copying blog content 1:1

• no links to products or services

• incorrect DNS configuration

• changing the main domain record instead of the subdomain

• outdated content

• buttons leading to the wrong places

• custom HTML with errors

• publishing articles without checking them

• treating the knowledge base only as a page for bots

The knowledge base is public. Don’t publish test, incorrect, outdated, or copy-pasted content from other places.

What to do after configuration

After launching the knowledge base, do a few extra things.

Check:

• if DNS was configured correctly

• if Semly confirmed the activation of the subdomain

• if the knowledge base homepage works

• if the generated articles are visible

• if the sitemap works correctly

• if llms.txt works correctly

• if the buttons lead to the right pages

• if the knowledge base is linked from the homepage

• if the knowledge base was added to the XML sitemap

• if the articles contain links to products or services

Then keep an eye on:

• visibility of prompts related to articles

• AI answer sources

• brand position

• sentiment

• Semly recommendations

• competitor mentions

Related articles

• How to generate your first article

• How to read Semly recommendations

• What AI answer sources are

• How to create prompts and topics

• How to analyze prompt details

• How to add products or a Google Shopping XML file

Summary

A knowledge base on a subdomain lets you quickly publish organized content about the brand, products, and services without rebuilding the main site.

Key benefits:

• runs on Semly infrastructure

• is optimized for LLM bots

• automatically generates a sitemap

• automatically generates llms.txt

• supports structured data

• shares content in HTML, Markdown, and JSON-LD

• can expose public API and MCP

• lets you publish unique articles

• makes it possible to link to a landing page

• lets you edit buttons and add your own HTML

• is available to users and AI systems

After launching the knowledge base, add a link to it from the main page and include it in the XML sitemap. This helps users and bots find new content faster.