Finalist
Ferg and Kert
Massive agent swarm for automated Prospect origination
Core trick
Parallel search agents query Tavily for each brief, and every extracted prospect finding stores its source URL before entering the graph.
Description
Massive agent swarm for automated Prospect origination
README excerpt
<div align="center"> <img src="docs/media/banner.svg" alt="Growth Swarm" width="100%"> <br> [](src/) [](tsconfig.json) [](supabase/migrations/) [](src/lib/connectors.server.ts) [](src/components/GraphCanvas.tsx) [](#evidence-first-by-design) **Give it a brief. A swarm of agents researches it, builds one shared, cited graph, votes on whether the job is done, and hands the result to your CRM.** [**▶ Watch the demo**](#-see-it-run) · [**Run the demo locally**](#run-the-demo-in-30-seconds) · [**How it works**](#how-it-works) · [**Product spec**](PRD.md) </div> --- Most "AI research" tools return a confident paragraph. Growth Swarm returns a graph you can check. You describe what you want found and what a finished answer looks like. Between 5 and 100 agents split the brief into tasks, search the web, qualify what they find against your criteria, record dated buying signals, and identify the people who own the problem. Every claim links to the exact passage