You go to events. You meet people. You follow up. Nobody does.
Luma shows you who attended. 50, 100, 200 people. Their names. Their faces. All right there.
LinkedIn is where business happens. But manually searching each person? That's 3+ hours of work per event.
So nobody does it. The warmest leads you'll ever have — people you were literally in the same room with — go cold in 48 hours.
I automated the whole thing. Scan the guest list. Match LinkedIn profiles. Send personalized notes. Staggered. Human-paced. Automatic.
A Chrome extension that lives in your sidebar. A web dashboard to track every campaign. A Telegram bot watching over it all. Together, they turn attending an event into a stack of pending LinkedIn connections — with zero manual work.
Side panel UI. Detects Luma event pages automatically. Scans the full guest list, lazy-loading all attendees. Finds LinkedIn profiles. Queues connections. Runs entirely in your browser — no server required.
Track every campaign across every event. See pending, sent, accepted, and failed connections. Cumulative progress graph. CSV export. Real-time Supabase sync so the data is always current.
24/7 VPS monitoring. Responds to /status, /health, /logs commands. Pings on errors. UptimeRobot checks every 5 minutes. Also routes queries to Claude with MCP access for intelligent debugging.
Four steps. The first one is manual. The rest run on their own.
Warm intros are the #1 pipeline channel at every early-stage company. Most people never capture them. Building the infrastructure to do it at scale meant solving a few genuinely hard problems.
LinkedIn's public API requires business approval and doesn't allow connection requests. So I reverse-engineered their internal Voyager REST API — the same one their own frontend uses.
The extension extracts the CSRF token from the JSESSIONID cookie, crafts the right headers (x-li-track, x-li-page-instance), and calls the
voyagerRelationshipsDashMemberRelationships endpoint directly. All routed through a hidden LinkedIn tab to stay same-origin and avoid CORS blocks.
Aggressive outreach kills relationships. So this is deliberately slow: 25 connections/day, 15-30 min random delays, only during 6am-11pm local time. Zero bans in production. The insight here is a GTM one, not an engineering one: volume without trust is noise.
GTM operator turned builder. I hate doing things manually that a computer can do for me. So I build the tools. Targeting GTM + Partnerships roles at Series A-B AI companies.