Live discovery feed. Refreshed hourly. REST + MCP.

New Startups API

A continuously refreshed feed of newly discovered AI startups, found across the entire public web by predictive AI and verified before they go live. Sort, filter, paginate. Built for sales teams, investors, and agents that need the freshest data.

Get Free API Key Full API Reference
curl
GET /api/v1/startups
curl -G https://www.startuphub.ai/api/v1/startups \
  -H "Authorization: Bearer sk_live_..." \
  --data-urlencode "sort=created_at.desc" \
  --data-urlencode "sector=AI Agents" \
  --data-urlencode "country=US" \
  --data-urlencode "founded_after=2024-01-01" \
  --data-urlencode "limit=100"
9New in last 24h
187New in last 7d
2,608New in last 30d
21,440Total in registry

Five most recent additions

Pulled from the same endpoint your API call hits. Refreshed every 5 minutes.

See the full feed
NameAdded
Transportation Security Administration11h ago
United States Immigration and Customs Enforcement11h ago
Universo Online11h ago
Jadely12h ago
Seo Mcp AI12h ago

How it works

Three calls. JSON in, JSON out. No SDK to install.

1

Pull the freshest first

GET /api/v1/startups?sort=created_at.desc&limit=100 with your Bearer key. Returns the 100 most recently added startups, JSON in, JSON out.

2

Filter for your ICP

Layer in &sector=AI Agents, &country=US, &employees_min=2&employees_max=50, &founded_after=2024-01-01. Combine any of them. Stealth-mode rows hidden on free / Pro Lite.

3

Poll daily, or page through history

Cursor with &offset= for backfills. For ongoing freshness, store the highest created_at from the last call and use it as your watermark — only paginate when new rows appear above it.

Drop into any stack

Plain HTTPS + JSON. No SDK required.

Node.js / TypeScript
const params = new URLSearchParams({
  sort: 'created_at.desc',
  sector: 'AI Agents',
  country: 'US',
  founded_after: '2024-01-01',
  limit: '100',
});
const r = await fetch(`https://www.startuphub.ai/api/v1/startups?${params}`, {
  headers: { Authorization: 'Bearer sk_live_...' },
});
const { data, pagination } = await r.json();
// data is an array of startup rows, freshest first.
Python
import requests

r = requests.get(
    'https://www.startuphub.ai/api/v1/startups',
    headers={'Authorization': 'Bearer sk_live_...'},
    params={
        'sort':          'created_at.desc',
        'sector':        'AI Agents',
        'country':       'US',
        'founded_after': '2024-01-01',
        'limit':         100,
    },
)
data = r.json()
for s in data['data']:
    print(s['name'], '—', s['one_liner'])
Response
{
  "data": [
    {
      "id":              "uuid",
      "name":            "Acme AI",
      "slug":            "acme-ai",
      "one_liner":       "Voice agents that close on the first call.",
      "website":         "https://acme.ai",
      "sectors":         ["AI Agents", "Voice AI"],
      "hq_country":      "US",
      "hq_city":         "San Francisco",
      "employee_count":  12,
      "total_funding":   3500000,
      "founded_date":    "2025-08-14",
      "created_at":      "2026-05-12T09:32:11.142Z",
      "total_score":     67
    }
  ],
  "pagination": { "total": 13427, "limit": 100, "offset": 0 },
  "credits":    { "used": 1, "limit": 100, "remaining": 99 }
}

What’s in the box

Every column, every filter, every channel.

Tapped into the entire public web

Other directories scrape each other and lag. We continuously scan the entire public internet and every public data source, then run it through proprietary predictive AI that surfaces a company early and verifies it against its own website before it ever goes live.

Sort + filter at the column level

sort by created_at, founded_date, total_funding, employee_count, employee_growth_quarterly_cagr, latest_funding_date, or total_score. Filter on sector (jsonb match), hq_country, operating_status, employees range, founded date range.

Stealth-mode visibility

Pro and above unlock startups marked stealth — companies our predictive AI surfaces before they have a public website. The richest source of pre-revenue deal flow you can get programmatically.

Full row, every column

Each row returns: name, slug, website, description, sectors, hq_country, hq_city, employee_count + growth, total_funding, latest_funding_date, founder_uuids, total_score, and the enrichment metadata. Drop straight into your CRM, BI tool, or scoring pipeline.

MCP server for Claude / Cursor

Same endpoint exposed as the search_startups MCP tool. "Show me AI agent startups in the US founded after 2024 with under 20 employees" — one natural-language query, structured rows back.

Cache-friendly + paginated

Stable cursors on (created_at, id), HTTP cache headers honour Cache-Control, rate-limit headers expose how many credits remain in this window. Drop-in for any sync pipeline.

Built for these workflows

Real production patterns from teams using the API today.

Sales prospecting

Run a daily cron against the API for your ICP filter. Push net-new rows into your CRM with a "fresh — never contacted" tag. Outbound to companies before competitors find them.

Investor deal flow

Configure your screening criteria once. Get every newly discovered match in your sector + geography the day it lands. The earliest signal anywhere outside a private network.

Competitive intel

Watch your sector for new entrants. Get an alert the moment a new company appears with overlapping keywords. Stay ahead of the press cycle.

AI agent workflows

Connect via MCP to Claude / Cursor / Windsurf. "Find AI agent startups founded in the last 90 days with 5-25 employees in Europe." One query, ranked rows back, ready to score or contact.

vs. the incumbents

Fresher data. Lower cost. Agent-ready out of the box.

ProviderTheir offeringStartupHub.ai New Startups API
Crunchbase APIEnterprise sales cycle. $$$ contract. Refreshed weekly.
Self-serve key in 30 seconds. Daily refresh. Stealth-mode rows incumbents don't have.
PitchBook / CB InsightsFive-figure annual contracts. No agent / MCP integration.
API-first. MCP-native. Pay for what you query, not a seat license.
LinkedIn Sales NavNo API. Manual search only. Saved-search emails delayed.
Real API. Daily-fresh rows. Programmable filters.
Apollo / ZoomInfoStrong on established companies, weak on net-new and pre-launch.
Built specifically for fresh-discovery use cases. Stealth + early-stage coverage.
Y Combinator company listYC-only. Public list. Two batches a year.
Every batch. Every accelerator. Plus non-accelerator companies. 24h cadence.
MCP Server

Search the registry from inside Claude.

The same New Startups API ships as a Model Context Protocol server. One URL into your AI client and your assistant gains the search_startups tool — natural-language sourcing without leaving the chat. “Find AI agent startups in Europe founded in the last 6 months with under 25 employees.”

Claude DesktopClaude.ai webCursorWindsurfClaude Code
See MCP setup
claude_desktop_config.json
{
  "mcpServers": {
    "startuphub": {
      "url": "https://www.startuphub.ai/api/v1/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_..."
      }
    }
  }
}

Frequently asked questions

Start sourcing fresh AI startups today

Free tier, no credit card. 30 seconds to your first API call.

Get Free API Key