Regular AI chatbot
- Only knows what it learned in the past
- Has never seen your private files
- Can sound sure even when it is wrong
- Does not tell you where the answer came from
What is RAG?
Look it up first, then answer
RAG stands for Retrieval Augmented Generation. That sounds fancy, but the idea is simple. Before the AI answers, it looks up the answer in your documents. Then it writes the reply using what it found.
Think of an open book test
A regular AI chatbot is like a student taking a test from memory. Most of the time it does fine, but sometimes it guesses and gets things wrong.
RAG is like letting that student bring your textbook. Before answering, the student flips to the right page, reads it, and then writes the answer. They can even tell you which page they used.
How skimeasy does it, step by step
Here is what happens behind the scenes. You don't have to do any of this. We do it for you.
1 Cut it into small pieces
Long documents are hard to search. So we cut each one into small pieces, about a paragraph each. These pieces are called chunks.
2 Give each piece a "meaning tag"
Next, we turn each chunk into a list of numbers that describes what it is about. This is called an embedding. Pieces about similar topics get similar numbers, even if they use different words. For example, "car" and "vehicle" end up close together.
We save these in a vector database, which is just a smart filing cabinet for these numbers.
3 Find the best matches
When someone asks a question, we give the question its own meaning tag too. Then we look for the chunks with the closest match. It is like finding the right pages in a book without reading the whole thing.
4 Write a clear answer
We hand those few chunks to the AI and say: "Answer using only this." The AI writes a short, clear reply and tells you which file it came from.
If none of the chunks match, the assistant says it could not find the answer. No guessing.
Why does this matter?
Your skimeasy assistant
- Knows your newest documents
- Answers only from your files
- Says "I don't know" instead of guessing
- Shows the file and page for every answer
Word helper
Some tech words you might hear, in plain English.
- AI model
- A computer program that can read and write like a person. ChatGPT is one example.
- Chunk
- A small piece of a document, about a paragraph long.
- Embedding
- A list of numbers that describes what a chunk is about. Chunks about similar things get similar numbers.
- Vector database
- A special filing cabinet that stores those numbers and finds the closest match really fast.
- Iframe
- A little window on a web page that shows something from another website, like your assistant.