YBOT: Building an AI That Actually Knows Your Website
For years, websites have had search boxes.
You type in a question. The website gives you a list of links.
Then you click through page after page trying to find the answer.
YBOT changes that. YBOT isn’t just a chatbot bolted onto a website. It’s an attempt to create something closer to a digital guide—an AI that has actually read the website, remembers what’s on it, and can have a conversation about it.
Step One: Teaching the AI Your Website
The first problem with AI is that it knows a lot about the world, but very little about your world. Your website may have ten years of articles, opinions, research, and stories that no public AI model has ever seen.
So YBOT builds its own memory.
Instead of reading every article every time someone asks a question, YBOT creates an index. Think of it like a library catalog.
Every post and page is:
Read
Cleaned of HTML and formatting
Broken into manageable chunks
Indexed into a database
This gives the AI a searchable memory of the website.
Step Two: The AI Becomes an Interpreter
When someone asks: “What have you written about USAID?”
YBOT doesn’t search Google. It searches its own memory.
It finds the relevant articles, sends them to the language model as context, and then lets the AI explain the information in plain English.
The result feels less like a search engine and more like asking someone who has read the entire website.
Step Three: Showing Its Sources
One of the biggest problems with AI is that it often sounds convincing, even when it’s wrong. So YBOT does something different.
Whenever possible, it shows its work.
Ask a question about USAID, inflation, AI, or Cuba, and YBOT can say:
“Here are the articles this answer came from.”
Then it provides clickable links directly to those posts.
The AI becomes:
A guide
A librarian
A researcher
Not just a talking machine.
Step Four: Giving It a Personality
Most chatbots sound like they graduated from Corporate Customer Service University. YBOT doesn’t have to.
Each website can give YBOT its own personality:
Serious
Humorous
Sarcastic
Professional
Like an old grandfather explaining things to his adult grandchild
The personality is simply another layer of context.
The same engine can become:
An Alien commentator
A business consultant
A historian
A teacher
A news analyst
Step Five: A Living Memory
Large language models eventually become outdated.
The world changes. Presidents change. Wars begin. Companies fail. Technologies emerge.
So YBOT has a second brain.
A persistent memory database.
It stores:
Current facts
Site-specific information
Important updates
Knowledge collected by external Python agents and scrapers
This means the AI can know things that happened this morning, even if the underlying model was trained months ago.
Step Six: The Hybrid Architecture
Under the hood, YBOT is actually several systems working together.
Website Content
↓
Content Index
↓
Memory Database
↓
External Knowledge Feed
↓
AI Model (Ollama, Llama, Qwen, etc.)
↓
YBOT
And that’s the real idea. The AI isn’t the product. The context is the product. The memory is the product.
The organization of information is the product.
The language model is simply the interpreter.
YBOT is a WordPress plugin that acts as the user interface for a local AI system. When a visitor asks a question, the plugin first searches a MariaDB database that contains indexed website content, persistent memory, and optionally external knowledge collected by Python agents. The plugin then builds a context package and sends it to an Ollama server running local language models such as Llama or Qwen. The AI generates a response using both the model's knowledge and the website's indexed information, and the answer is returned to WordPress, where YBOT displays the response and can also provide links to related articles on the site. And yes, it was written by Ai in less than 2 hours.
Why This Matters
The internet is drowning in information. People don’t need another search box.
They need an intelligent guide that understands the knowledge already sitting on a website and can help them navigate it.
That is what YBOT is trying to become. Not an artificial intelligence.
But an artificial memory and interpreter for human knowledge.
