The Emerging VC reposted this
Here is a step-by-step and signal-by-signal guide to building your own AI-powered deal sourcing engine from scratch. Pre-2022: VC deals came either inbound or outbound. Post-2022: VC deals come either inbound, outbound or from AI-detected signals. Deal sourcing is becoming a technological competition in the race to find and connect with top startups. Firms are using AI to scrape and filter data that will point them to new relevant deals for their firms. If you are an emerging VC, you want to be thinking about how you can use technology to stay ahead in your deal sourcing capabilities. That's why I am giving you the blueprint to help you build your own AI-powered deal sourcing engine from scratch. Step 1: Connect to signals Here are 14 publicly available data points that indicate a founder is forming or a startup is emerging. These fall within 5 broad categories. 👉 Formation signals → New company registration → Stealth role added to professional profile → Position departure with no announced next step 👉 Intellectual property signals → New patent filing → Grant awarded → Research paper published 👉 Builder signals → New GitHub repository opened → GitHub activity spike → Hackathon finalists → Product Hunt launch → App store launch 👉 Intent signals → Co-founder search posted publicly → Domain registration 👉 Social graph signals → Investor interaction detected Step 2: Triangulate to separate noise from signal There is not one single signal you can track which will surface all the best deals for you. You need to be tracking multiple. Think about it, a patent filing could be defensive. A domain registration could be a side project. What you are looking for is a cluster: multiple signals from the same person or entity in a short window. Step 3: Filter for thesis fit Once you have a list of active signals and deals start trickling in, you need to run them through your firm’s investment criteria i.e. the stage, sector, geography and/or founder profile your fund invests in. That way you are separating relevant signal and relevancy for your firm from noise. How long does this take to build? A useful internal version tracking enough of the above signals could take 3–6 months depending on the level of time, skill and resource is available within your VC firm. A broader system covering most of these signals could take 6–12 months, with ongoing maintenance and data costs. So you can definitely build this. But it's a lot of work. If you are up for the challenge great, otherwise, Evertrace have already built all of this. Their entire focus is surfacing deals faster than anyone else, and they are regularly adding new signal types to the platform. Check them out.