Front Run

The problem

When a startup raises money, a short, high-intent window opens — they're about to hire and spend, and whoever reaches them first wins. But by the time a round hits the news, every sales team is already piling in. Speed-to-lead is the whole game, so I asked: what if you could catch the raise the day it happens?

What I built

• Built an autonomous AI SDR — an “AI employee,” not a chatbot — that scrapes SEC Form D filings straight from EDGAR the same day they post, catching a raise on day zero, before the news breaks or rival agencies even know it happened.
• From that single signal it runs the whole top of funnel itself: researches the company, finds the right contact, drafts a tailored email, and sends it — no human in the loop.
• Handles the conversation, not just the first touch: a multi-agent system reads each inbound reply, classifies intent (interested / unsure / not), drafts the next move, and books the meeting itself via Cal.com.
• I owned the outreach + reply loop — sending, multi-agent reply-triage, and booking — then integrated all four teammates' workstreams into one deployable full-stack Next.js app and owned deployment.
• Shipped end-to-end in a ~8-hour build with a team of 4.

Impact

• Turns a free, public, instant signal (EDGAR) into a same-day outreach edge — the moat isn't secret data, it's being first to act on data everyone can see. Speed-to-lead as a product.
• Replaces the manual SDR grind as a closed loop: detect, research, write, send, triage, book — a self-running employee, not another lead list or reply-suggestion tool.
• My piece — the reply loop plus the integration and deployment work — is what made four separate hackathon pieces cohere into one product that actually runs.