Developer Preview - Try the FREE API now

Knowledge Base
for your AI

Upload docs. Ask questions. Get answers with sources.
Even images and videos inside your docs are understood by AI.

Get Early Access See the Demo ↓

Works with any AI model - Connect via tool call or MCP · OpenAI · Anthropic · Gemini · Any LLM

Your AI + Ragionex in action

Your AI → Calls Ragionex to Ask Questions in Natural Language → Gets Answers with Sources

Uses Ragionex as a tool · Multi-query search for complex topics · Answers questions about images & videos in results

Your AI AssistantRole: VS Code Expert

Ask a question...

How it works

Three steps to AI-ready docs.

Upload your docs

PDF, DOCX, Markdown, or any documentation format.
We handle the rest.

We process everything

Docs, images, and videos are analyzed and optimized for search. Fully automatic.

Your AI asks, we answer

Connect via API, MCP, or tool call.
Your AI gets the answer with its source. Every time.

Why existing RAG fails

And how Ragionex fixes it.

Traditional RAG

Calls LLM on every query

Every search = API call to GPT/Claude. Costs add up fast.
Hallucinates

LLM generates answers and can invent facts that don't exist in your docs.
Low retrieval accuracy

Embeds raw text. If your phrasing doesn't match the chunk wording, it misses.
Ignores images & videos

Visual content in your docs is invisible to the search.
Slow

LLM inference on every query adds seconds of latency.

Ragionex

Zero runtime AI cost

No LLM call at search time. All AI work is done beforehand.
Zero hallucination

We don't generate answers. We find them in your docs. Can't make things up.
High retrieval accuracy

Multiple question aliases per answer. Your query matches regardless of how you phrase it. Every result cited.
Images & videos are searchable

Visual content in your docs is understood and included in search results.
Fast

Pure search, no LLM waiting. Results in milliseconds.

Dead simple API

One endpoint. One API key. That's it.

Try it now for FREE - Developer Preview

This demo API is trained on VS Code documentation. Ask any VS Code question to test it.

Why VS Code? Every developer knows it, it's open source so docs are publicly verifiable, and it's packed with images and videos - perfect for showcasing visual content search.

Your API Key

rgx_pk_preview_7f3a9b2c

VS Code docs licensed under CC-BY 3.0 US · Code samples under MIT License · Source: microsoft/vscode-docs

Request

import requests

response = requests.post(
    "https://api.ragionex.com/search",
    headers={"X-API-Key": "YOUR_API_KEY"},
    json={
        "question": "How to go to super implementation in Java?",
        "results": 10
    }
)

print(response.json())

const response = await fetch("https://api.ragionex.com/search", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "X-API-Key": "YOUR_API_KEY"
  },
  body: JSON.stringify({
    question: "How to go to super implementation in Java?",
    results: 10
  })
});

const { results } = await response.json();
results.forEach(r => console.log(r.answer));

curl -X POST https://api.ragionex.com/search \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{
    "question": "How to go to super implementation in Java?",
    "results": 10
  }'

                                    
                                    
                                    
                                
Response

                                {
  "success": true,
  "results": [
    {
      "answer": "## Go to Super Implementation

You can keep track of class implementations and overriding
methods by clicking the **Go to Super Implementation** link
on hover.

![Spring Navigation](https://media.githubusercontent.com/media/
microsoft/vscode-docs/.../images/java-editing/goto-super.png)
<media-description for="images/java-editing/goto-super.png">
This screen shows a Java code editor with a tooltip popup
displaying method documentation. EDITOR: public class PetTy
private final @Autowired public PetType this.pets @Override
public String print(PetType petType, Locale locale) {
return petType.getName(); } TOOLTIP: Go to Super
Implementation String org.springframework.samples.petclinic
.owner.PetTypeFormatter.print(PetType petType, Locale locale)
Print the object of type T for display. Parameters: petType
locale Returns: the printed text string.
</media-description>

<video src="https://media.githubusercontent.com/media/
microsoft/vscode-docs/.../images/java-editing/
goto-super-implementation.mp4" autoplay loop muted
playsinline controls title="Go to super implementation">
</video>
<media-description for="images/java-editing/
goto-super-implementation.mp4">
## PHASE 1: INITIAL STATE

Dark-themed Visual Studio Code window displaying the file
AdvancedPetValidator.java. The main content area shows:
- Line 6: public class AdvancedPetValidator extends PetValidator {
- Line 8:     @Override
- Line 9:     public boolean supports(Class<?> clazz) {
- Line 10:        return true;

## PHASE 2: EVENTS

EVENT 1:
- WHAT: Line 10 containing return true;
- WHERE: Code editor, line 10
- ACTION: Hover
- RESULT: A tooltip appears below line 10. The tooltip
  contains: "Go to Super Implementation" boolean
  org.springframework...supports(Class<?> clazz)

EVENT 2:
- WHAT: "Go to Super Implementation" link in tooltip
- WHERE: Tooltip below line 10
- ACTION: Click
- RESULT: The editor switches to the file PetValidator.java.
  Line 59: public boolean supports(Class<?> clazz) {
  Line 60:     return Pet.class.isAssignableFrom(clazz);
</media-description>",
      "source": "https://raw.githubusercontent.com/microsoft/vscode-docs/.../docs/java/java-editing.md",
      "id": "F7VBU2REK6"
    },
    // ... 9 more unique, LLM-ready results with sources
  ]
}
                            

Get early access

Be first to try Ragionex with your own docs.

We're opening Closed Beta soon. Drop your email and we'll notify you when it's your turn.

No spam. Just one email when beta opens.

Knowledge Base for your AI