University IP Scout

• 

  Telynx terminal output simulation

Inspiration

VCs and researchers waste hours manually reviewing university tech transfer offices for relevant intellectual property. Most IPs are irrelevant to their focus areas, resulting in:

Wasted time reviewing hundreds of irrelevant technologies High API costs scraping everything Missed opportunities in the noise

What it does

Agent S solves this: Automatically scout, classify, and filter IPs from multiple universities in minutes, not hours.

How we built it

Universal URL Extraction Works across MIT, Stanford, HKU, NUS, and 50+ universities Handles pagination, "Show More" buttons, and dynamic content Auto-translates non-English content (Chinese, Japanese, etc.) LLM Pre-Filtering Classifies IPs BEFORE scraping (saves 70% time & cost) Categories: Life Sciences, Medical Devices, Other Smart hybrid mode: Heuristics → LLM for uncertain cases Hybrid Scraping Traditional scraping (BeautifulSoup) for speed OpenAI/AGI web search tool fallback for JS-heavy/blocked sites Automatic quality detection and retry logic

Challenges we ran into

With Pecking university, where the website has a search bar for manual IP search.

What's next for University IP Scout

Test and scale with more universities.

Built With

• agi
• intel-galileo
• lovable
• openai
• python-package-index
• redis
• telnyx