/web-crawl /deepfake-detect /llm-report /cluster
OSINT PLATFORM iDEX ADITI 4.0 · PS-18

Anveshak

अन्वेषक — Eyes Across the Open Web.
AI-powered OSINT monitoring, media verification, and local LLM reporting. Runs on one machine. No internet required.

Request Evaluation ← All Products
M1 · SOURCE CREDIBILITY · M2 · WEB CRAWLER · M3 · SOCIAL · M4 · VISION · M5 · REPORTER
// FIVE MODULES

What Anveshak Does

MODULE M1

Source Credibility

Automated credibility scoring with analyst feedback loop. Every score change is audit-logged — court-admissible chain of custody for source reliability.

MODULE M2

Web Intelligence

Crawl4AI-powered open-web crawling with multilingual NLP, entity extraction, and narrative clustering. Monitors defined topics continuously.

MODULE M3

Social Monitoring

Real-time adapters: Telegram, Reddit, Bluesky, X/Twitter. Configurable signal thresholds fire alerts when narrative clusters breach your watch criteria.

MODULES M4 + M5

Vision + Reports

YOLOv8 object detection, CLIP similarity, deepfake probability scoring (float 0–1, not boolean), EXIF forensics. LLM report generation via local Ollama — data never leaves.

// OPERATIONS

Field Use Cases

USE CASE 01

Narrative Tracking

A coordinated disinformation campaign surfaces across Telegram and Reddit. Anveshak clusters narratives by independent source count, fires a signal when the threshold is breached, and generates a sourced LLM report — fully offline.

USE CASE 02

Media Verification

A video claiming to show an IAF incident circulates online. Anveshak's M4 module runs YOLO detection, deepfake probability scoring, and EXIF forensics — delivering a verifiable technical assessment before the news cycle moves.

USE CASE 03

Watch Topic Monitoring

An analyst defines a watch topic — keywords, platforms, signal threshold. Anveshak monitors continuously, fires WebSocket alerts to connected sessions in real-time, and maintains an auditable credibility log for every source.

// UNDER THE HOOD

Sovereign by Design

All LLM inference runs on Ollama — localhost only. Intelligence data never leaves the deployment boundary.

Python 3.12 FastAPI + Pydantic v2 PostgreSQL + pgvector Redis + ARQ Ollama (local LLM) YOLOv8 + CLIP spaCy 3 (multilingual) Crawl4AI Telethon / PRAW / atproto React + MapLibre GL Docker Compose / k3s
// THE SUITE

Scale When Ready

ENTRY
Grosint
→
YOU ARE HERE
Anveshak
→
FUSION
Drishti
// ENGAGE

Request an Anveshak Briefing

Built for iDEX ADITI 4.0 PS-18 — reach us directly on WhatsApp.

Message us on WhatsApp

+91 9901938800