> ## Documentation Index
> Fetch the complete documentation index at: https://docs.revisory.app/llms.txt
> Use this file to discover all available pages before exploring further.

# AI transparency

> How Revisory uses AI and what happens with your data.

<Note>
  🇩🇪 [Auf Deutsch lesen](/de/datenschutz/ki-transparenz)
</Note>

Revisory uses AI for document analysis. Here's how it works and what control you have.

## How is AI used?

AI is used in two areas:

### 1. Text extraction

After upload, AI extracts text from your documents:

* PDF text recognition (OCR)
* Structure recognition in tables
* Conversion to a unified format

### 2. Evaluation

During analysis, AI evaluates each document against your defined evaluation steps:

* Comparison with criteria
* Scoring according to the defined scheme
* Feedback generation

## Which AI model is used?

Revisory uses modern language models (Large Language Models). The specific models may change to ensure the best quality.

## Is my data used for training?

**No.** Your documents and analyses are not used to train AI models. The AI only processes your data to answer your requests.

## Data processing by AI

When you start an analysis:

1. The extracted text is sent to the AI system
2. The AI system evaluates the text
3. Results are stored
4. The sent text is not stored or reused

## Limitations of AI

Be aware of the limitations:

* **Not perfect:** AI can make mistakes
* **Context:** Highly specialized terminology may not be recognized
* **Interpretation:** Evaluation depends on the quality of your criteria definitions

<Tip>
  The more precisely your evaluation steps are formulated, the more meaningful the AI evaluations become.
</Tip>

## Human control

AI is a support tool, not a replacement for human judgment:

* Review results critically
* Adjust evaluation steps if results don't seem sensible
* Use AI feedback as a starting point for your own assessments