DEV Community

Cover image for I Built an AI Startup Radar: 1100+ YC Companies on One Screen (2025 Trends Included)
Vastargazing
Vastargazing

Posted on

I Built an AI Startup Radar: 1100+ YC Companies on One Screen (2025 Trends Included)

I Built an AI Startup Radar: 1100+ YC Companies on One Screen (2025 Trends Included)

I spent the last 2 months of evenings scraping and analyzing every Y Combinator AI startup from W22 to W25.

Live demo: https://yc-ai-radar.vercel.app/

Why I built it
YC has become almost a pure AI incubator:

  • 2023: 44% AI startups
  • 2024: 72% AI startups
  • 2025: around 80% AI startups

Everyone keeps asking the same questions:

  • Which niches are already overcrowded?
  • Where is YC actually betting money in 2025?
  • Where should I definitely NOT build to avoid becoming the 152nd clone?

Most solutions out there are Excel files, random tweets, or hours of manual digging. I wanted one single place that shows the real, normalized picture of the market.

What the tool does

  1. Explore page – 1100+ startups with smart filters
    • Categories (Agents, DevTools, Healthcare, Legal, Infra, etc.)
    • Crowdedness level (Few / Some / Many players)
    • Batch filters (W22 – W25)
    • B2B vs B2C
    • One-liner + direct link for every company

  2. Trends page – market share growth instead of raw counts
    YC changes batch sizes every season, so absolute numbers lie.

Example:
S24: 10 crypto out of 200 = 5%
W25: 8 crypto out of 100 = 8%
Raw count looks like decline → actually +60% market share growth.

  1. Opportunities page – gap analysis + high-barrier warnings Legal, Healthcare, Finance look empty from afar… until you remember 12-18 month sales cycles, FDA approvals, heavy regulation, etc.

Biggest 2025 insights I found

  • AI Agents: from 7% (W22) to 29.2% (W25) = +157% market share
  • Education exploded: +371% market share
  • Enterprise Ops is dying: –54%
  • Legal cooled off hard: –60% (looks like YC finally felt the barriers)
  • B2B dominates completely: 85% of all AI startups (B2C only 11%)

Tech stack & key decisions

  • Python for scraping and analytics
  • Rule-based categorization (yes, not LLM – reason below)
  • React + Tailwind + Vercel frontend

Why rule-based instead of LLM:

  • 100% deterministic (same input = same output forever)
  • Fully explainable
  • Processes 1100+ startups in under 1 second
  • Zero hallucinations or drift

Who this is for

  • Founders & indie hackers → validate your idea in 30 seconds
  • Investors → instantly see where YC is putting money right now
  • Job seekers → find the fastest-growing companies

Links
Live demo: https://yc-ai-radar.vercel.app/
CSV dataset – download button right on the site
GitHub – coming after a quick refactor
100% free, no login, no nonsense.

Go see how many clones of your idea already exist in YC.

Feedback and upvotes are very welcome!

Top comments (0)