Api Hub / Sections / Section Upload Methods

RAG Documents & Video — Upload and Analysis

Upload documents, images or videos and query them with AI through the single universal endpoint.

📡 SINGLE ENDPOINT FOR ALL FUNCTIONS

https://app.neuralaiweb.com/functions/neuralaiPublicApi

All functions — chat, image, video, music, document RAG, video RAG — go through this single POST endpoint. The orchestrator automatically detects intent from the prompt or you can force it with force_intent.

Part 1 — Document RAG

🔄 Document RAG Pipeline

File upload → https://app.neuralaiweb.com/functions/publicUploadFile
Text extraction / OCR → https://app.neuralaiweb.com/functions/extractDocumentText
Inject text into prompt → chat con https://app.neuralaiweb.com/functions/neuralaiPublicApi

📄 Supported document formats (native OCR)

PDF — including scanned PDFs (native OCR)
DOCX / ODT — Word and LibreOffice documents
XLS / XLSX — Excel spreadsheets
PPT / PPTX — PowerPoint presentations

TXT / MD / CSV / JSON / HTML — plain text
PNG / JPG / JPEG / WEBP / GIF / BMP — images with visual OCR
Maximum 3 files per request
Size limits: text files (TXT/MD/CSV/JSON/HTML) 50MB · PDF/DOCX and images 20MB · audio 30MB · video 150MB

Step 1 — File upload (public endpoint without auth)

Use https://app.neuralaiweb.com/functions/publicUploadFile to upload the file and get a permanent URL.

javascript

const UPLOAD_URL = 'https://app.neuralaiweb.com/functions/publicUploadFile';

async function uploadFile(file) {
  const formData = new FormData();
  formData.append('file', file);

Step 2 — Text extraction / OCR

Extract text content from the file (scanned PDF, images, Office, etc.) via OCR.

javascript

const EXTRACT_URL = 'https://app.neuralaiweb.com/functions/extractDocumentText';

async function extractText(fileUrl, fileName) {
  const res = await fetch(EXTRACT_URL, {
    method: 'POST',

Step 3 — RAG Chat with the document

Inject the extracted text into the prompt and send it to the main endpoint.

javascript

const API_URL = 'https://app.neuralaiweb.com/functions/neuralaiPublicApi';

async function ragDocumentChat(apiKey, userQuestion, documentText) {
  // Costruisci il prompt con il testo del documento
  const fullPrompt = `Hai accesso al seguente documento:

Complete example — Document RAG from scratch

javascript

const API_KEY  = 'nai_...';
const API_URL    = 'https://app.neuralaiweb.com/functions/neuralaiPublicApi';
const UPLOAD_URL = 'https://app.neuralaiweb.com/functions/publicUploadFile';
const EXTRACT_URL= 'https://app.neuralaiweb.com/functions/extractDocumentText';

Example — RAG with multiple documents (max 3)

javascript

async function chatWithMultipleDocs(files, userQuestion) {
  // Upload + OCR di tutti i file in parallelo
  const docs = await Promise.all(Array.from(files).slice(0, 3).map(async (file) => {
    const fd = new FormData();
    fd.append('file', file);

🧪 Pagina di Test Completa — RAG Documenti & Video

Copia questo file HTML completo, aprilo nel browser e testa subito il RAG documenti (con OCR) e il Video RAG senza installare nulla. Inserisci la tua API key, carica i file e fai domande.

📋 Instructions

Copy the code below and save it as test-rag-chat.html
Open it in any browser (double click on the file)
Enter your API Key in the top left field
RAG Documents — drag or select files, click Upload + OCR, wait, write your question
Video RAG — click on Video RAG tab, paste URL or upload file, write question, send

✅ Included features

• PDF, DOCX, ODT, XLS, PPT, TXT, CSV, HTML• Immagini con OCR visivo (PNG/JPG/WEBP)• Video RAG via URL pubblico• Video RAG via upload locale• Console di log dettagliata in tempo reale• TTS audio automatico nelle risposte

🧪Open RAG Test Page↗

Open directly in browser — no installation required.

html

<!DOCTYPE html>
<html lang="it">
<head>
  <meta charset="UTF-8" />
  <meta name="viewport" content="width=device-width, initial-scale=1.0" />

Part 2 — Video RAG

🎬 How Video RAG works

The video is analyzed frame by frame with audio and visual understanding via our LLM Orchestrator, and the result is returned as text. You can ask questions about the content, transcriptions, summaries, specific scenes, etc.

The video is downloaded/uploaded to the LLM Orchestrator
The LLM Orchestrator processes the video (takes 5–60s for long videos)
Your question is answered based on the video content

🎥 Supported video formats and limits

Accepted formats

MP4 (recommended)
MOV — Apple QuickTime
AVI — Audio Video Interleave
MKV — Matroska
WebM — web format

Limits

Maximum duration: 60 minutes
Max size: ~1 GB (via public URL)
Upload size: ~50 MB (via publicUploadFile)
Cost: 20 tokens per analysis

Mode A — Video RAG from public URL

If the video is already on a public server (YouTube direct, CDN, etc.), pass the URL in the video_url field. The server will download it and analyze it via LLM Orchestrator.

javascript

const response = await fetch('https://app.neuralaiweb.com/functions/neuralaiPublicApi', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    api_key: 'nai_...',

Mode B — Video RAG from local upload (user file)

If the video is local (user file input), upload it first with publicUploadFile, then pass the resulting file_url in the video_file_url field.

javascript

const UPLOAD_URL = 'https://app.neuralaiweb.com/functions/publicUploadFile';
const API_URL    = 'https://app.neuralaiweb.com/functions/neuralaiPublicApi';

async function videoRAGFromFile(videoFile, userQuestion) {
  // Step 1: Upload del video

💡 Example questions for Video RAG

"Summarize the video content in 5 key points"
"Transcribe all spoken dialogues in the video"
"Describe what happens at minute 2:30"
"Extract all text written on screen (video OCR)"
"How many people appear in the video and what do they do?"
"Is this video appropriate for children?"

Complete example — UI with file input and URL

html

<!DOCTYPE html>
<html>
<head><meta charset="UTF-8"><title>Video RAG NeuralAI</title></head>
<body>
  <h2>🎬 Video RAG</h2>

⚠️ Video RAG Technical Notes

Processing takes 10–60 seconds depending on video length
Video > 50 MB: use Mode A (public URL) — the server downloads directly
Video < 50 MB: use Mode B (upload) for maximum reliability
Private or authentication-required video URLs are not supported in Mode A
Uploaded videos are automatically deleted after analysis

📋 RAG field summary — POST request body

Field	Type	Required	Description
api_key	string	✅ Yes	Your API key (nai_...)
prompt	string	✅ Yes	Question or instruction for AI
force_intent	string	Recommended	'chat' for doc RAG · 'video_rag' for video
video_url	string	Video only (A)	Public URL of the video to analyze
video_file_url	string	Video only (B)	URL of video uploaded via publicUploadFile
file_urls	array	No	Array of document file URLs (max 3) — alternative to manual extraction

Deleting storage files from external apps

If your app generates or uploads files via the NeuralAI API key, you can list and delete them programmatically to free up your account's storage quota, by integrating these calls into your application's backend.

📡 Endpoint and requirements

https://app.neuralaiweb.com/functions/manageMediaStorage

API key with the 💾 Storage scope active, passed as x-api-key header (or Authorization Bearer)
Up to 1 GB everything works without tokens (free quota); beyond, each extra GB costs 50 tokens per month, charged automatically from the owner's wallet
The key can list and delete ONLY the key owner's files

Example — list and delete files from your app

javascript

const STORAGE_URL = 'https://app.neuralaiweb.com/functions/manageMediaStorage';

// 1) Elenca i file del tuo account storage (nomi, cartelle, dimensioni)
async function listMyFiles(apiKey) {
  const res = await fetch(STORAGE_URL, {

⚠️ Notes

Files still referenced by NeuralAI records (search history, document library, published sites) are skipped and counted in skipped
Deletion immediately frees up the key owner's storage quota
Calls must be integrated into your app's backend (the key is a secret: never expose it in frontend code)