Fluenta's idea pipeline is a two-layer system. This article covers layer one — scouting — the continuous scan that surfaces raw startup ideas. Layer two (verification) is covered in How Fluenta verifies ideas.

Why scouting is a separate layer

Most "AI idea generators" skip the scouting step entirely. They ask a language model to brainstorm. The output looks plausible but has no grounding in what's actually happening in the world.

Fluenta inverts the flow. Before any scoring, we observe. We read what the sharpest people in the market are already building, funding, writing about, and complaining about. The ideas come from the market, not from a prompt.

The 200+ sources we scan

Scouting draws from six source families, continuously:

1. Venture capital firms

What top-tier VCs and their portfolio companies are funding this week. Examples: a16z, Sequoia, Y Combinator, Kleiner Perkins, Lightspeed, Index Ventures, Greylock, General Catalyst, GGV, Pitango, and regional firms like Kalaari, Gobi Partners, Qiming, ZhenFund, Khazanah, Naspers, Launch Africa, Golden Gate Ventures, MAYA Capital, Upload Ventures.

2. Accelerators and ecosystems

Who got in, what they're building. Examples: Techstars, Station F, Plug and Play, Start-Up Chile, Seedcamp, SOSV, Endeavor, Penjana Kapital, Startup Nation Central, e27.

3. Top-tier business media

The trend-makers and trend-trackers. Examples: Forbes, Bloomberg, Wall Street Journal, Financial Times, Reuters, TechCrunch, Wired, The Economist, Fast Company.

4. Consulting and research

What the forecasting class is calling. Examples: McKinsey, BCG, Bain, Deloitte, PwC, EY, KPMG, Accenture, Roland Berger, Gartner, CB Insights, PitchBook, Harvard Business Review.

5. Academic and innovation

Where novel ideas enter the funnel. Examples: MIT, Stanford, and affiliated research groups.

6. Communities and discovery platforms

Where bottom-up demand leaks through. Examples: Product Hunt, GitHub, and public founder discussions.

How scouting actually runs

  1. Daily ingestion. Our crawlers and APIs pull new content from every source.
  2. Entity extraction. We extract companies, categories, pains, and tags using a purpose-built model.
  3. Deduplication and clustering. Similar ideas from different sources collapse into a single canonical entry.
  4. Provenance tagging. Every idea keeps a link back to every source that surfaced it — you can see the trail on the card.

What scouting does not do

Scouting surfaces candidates. It does not score them. An idea that shows up in Forbes and a16z on the same day is not automatically a winner — it still has to clear verification (layer two). That's where the Launch Readiness Score comes from.

What you see in the product

Every idea card in this week's ideas shows the scouting sources that surfaced it. The landing-page marquee is a live snapshot of the scouting roll.

What this means for you

You don't have to read 200 sources every morning. Fluenta does it for you, 7 days a week, and filters the noise down to the ideas worth scoring. The difference between scouting your own ideas (reading TechCrunch + a16z newsletters + 3 Reddit subs) and Fluenta's scouting layer is coverage and speed: we ingest from 200+ sources daily and deduplicate automatically. The ideas you see in the app are already clustered, sourced, and ready for verification — not raw links you have to sort through yourself.

Browse this week's ideas →

Related