yungbilionz/startup-os

StartupOS bills itself as an AI co-founder swarm. The pitch in the README is that a single startup idea is sent into the system, five specialized AI agents work on it at the same time, and roughly 60 to 90 seconds later the user gets back what the project calls an investor-ready package with a score out of 10. There is a live demo link on a Render-hosted URL. The five worker agents each have a specific role. A Market Researcher looks at market size, competitors, trends, and risks. A Business Analyst writes the revenue model, pricing, and go-to-market plan. A Pitch Writer produces the investor narrative including a one-liner, the problem, the solution, and the funding ask. A Financial Modeler generates three-year revenue, burn rate, lifetime value, customer acquisition cost, and margins. A Legal Scanner flags legal risks, licenses, intellectual property concerns, and compliance issues. A sixth Director Agent then reads all five outputs and writes the final synthesis along with the score. The project is built on the Swarms Framework using its MixtureOfAgents pattern, which is what lets the five specialists run in parallel and then hands the results to an aggregator. The language model behind every agent is LLaMA 3.3 70B served by Groq, which the README highlights as both fast and free. The backend is Flask, the frontend is plain HTML, CSS, and JavaScript, and configuration goes through a .env file using python-dotenv. Python 3.10 or newer is required. Setup involves cloning the repo, installing dependencies including swarms, flask, flask-cors, groq, python-dotenv, and google-genai, and signing up for two free API keys, one with Groq and one with Swarms. Once both keys are in the .env file inside the startup_os folder, you run python main.py and open localhost on port 5000. The README also documents a single REST endpoint, POST /analyze, that takes a JSON body with an idea field and returns market research, business model, pitch, financial, legal, and final synthesis sections. The README closes with an example for an app that connects African farmers to city buyers, reporting a 56 billion dollar total addressable market, a 5 percent transaction fee model, a 2 million Series A ask, and an 8.5 out of 10 score. The project is MIT-licensed.