About OpticParse & PhishVision

OpticParse and PhishVision were built by Paras Tejpal, a solo developer from Nurmahal, Punjab, India.

The idea came from a simple frustration: web scrapers break constantly. Every time a website updates its design, developers spend hours fixing selectors, XPaths, and CSS classes. There had to be a better way.

The answer was obvious in hindsight — stop parsing HTML and start looking at what humans actually see. If a human can read a webpage, an AI looking at a screenshot can too.

PhishVision came from the same insight applied to cybersecurity. Traditional phishing detection checks URLs against blacklists. But attackers create new domains every day. Visual detection — actually looking at what the page shows users — catches what blacklists miss.

• OpticParse: Vision-based web scraping API that ignores HTML entirely. If data is visible on screen, we can extract it — regardless of DOM structure, JavaScript rendering, or bot protection.
• PhishVision: Visual phishing detection API that analyzes page UI for brand impersonation, urgency tactics, and hidden prompt injection attacks targeting AI agents. Faster than funded competitors.

Tech stack: Python/FastAPI, Node.js/Express, Playwright, Browserless.io, Supabase, Cloudflare, 6-provider AI rotation (Groq, Gemini, GitHub Models, OpenRouter, DeepSeek, Mistral)

We are the only web scraping API that lets you run AI inference entirely on your own device using WebLLM.

When you enable local mode in your dashboard, the AI model downloads to your browser and runs on your GPU — no data leaves your machine, no API tokens consumed, complete privacy.

For teams that prefer cloud processing, our 6-provider rotation ensures maximum uptime with automatic failover across Groq, Gemini, GitHub Models, OpenRouter, DeepSeek, and Mistral.

Email: support@opticparse.com
Website: opticparse.com
Dashboard: dashboard.opticparse.com

For enterprise inquiries, partnerships, or grant-related questions, email us directly.