Mastermind · February 28, 2026
From a YC hackathon floor to a Hong Kong flagship
I spent the weekend on a hackathon floor at an accelerator's HQ, participating, not running, and the useful thing about being on the other side of the table is that you remember what the event feels like when the platform you're using isn't yours.
I hadn't been a participant in a while. I'd been running events, pitching events, sponsoring events, but not sitting at a folding table for forty-eight hours trying to build something with two strangers. It was a useful reset. A lot of the small frictions I'd been treating as acceptable in my own product, I noticed in other people's products the way a user would notice them, which is immediately and with annoyance.
The other reason the weekend mattered is that the whole thing was research for a much bigger event I'm planning. We're running what I think will be the biggest hackathon ever held in Hong Kong on July 3 and 4, a thousand participants, and I've been treating every event between now and then as rehearsal. Not marketing rehearsal. Product rehearsal. When a thousand builders use MentorMates end-to-end over a weekend, every failure mode surfaces, and the real test of the platform happens not in a pilot but under event-level load.
When I walked Xisen through this on the call, he pushed on whether I was confusing events with product milestones. It's a fair push. The trap would be treating the July event as a marketing beat, announcing it, promoting it, collecting logos, and then having the actual product underneath be the same as it was in February. That's not what the July event is for. The event is the forcing function. Every piece of the platform has to be good enough to survive it, and the question for every roadmap item between now and July is whether it's feeding the flagship or it isn't.
The workshop series I've been running in parallel became part of this on the call. Workshops were making real revenue, which mattered for the month, but I'd been treating them as a side hustle. Reframed properly, the workshops are a live research lane. They're where I watch new users onboard into the MentorMates ecosystem without the pressure of a live event, which means I get to see what breaks before it breaks at scale. The workshop feedback is effectively beta feedback for the flagship.
The partnership work this week was Hong Kong sponsor outreach. I've been pushing on three kinds of partners: partners who'd want to plug in as ecosystem sponsors, partners with distribution channels we don't have access to, and a local venue partner for the event itself. The frame I'm holding is the one Xisen and I landed on earlier this month: a partnership is only real if it converts into something concrete. A logo on a sponsor page isn't a partnership, it's a courtesy. A co-run workshop, a shared funnel, a defined deliverable, those are partnerships.
The sponsor deck conversation on the call was useful in the blunt way these conversations sometimes are. The deck is text-heavy, the structure doesn't tell a story, and it's trying to close three different kinds of reader at once. The fix isn't a polish pass, it's an audience cut. I need a deck for ecosystem sponsors, a different deck for distribution partners, and a third deck for venue and institutional partners. Each of those audiences is reading the same document looking for a different thing, and one document cannot serve all three.
The pattern I keep seeing in how I allocate time is that the flagship event either organizes the week or the week organizes itself around whatever inbound happened. The days I organize around Hong Kong produce compounding work. The days that get hijacked by the latest intro produce warm threads and no distribution. I said this out loud on the call because I wanted to name the pattern before it eats another month.
The browser-automation conversation came up on the floor this weekend too, and there's a real thread to pull on there for MentorMates, specifically around how an agent shows up inside an event to help organizers with the coordination work I've been trying to automate. That thread plugs directly into Xisen's agent-to-agent work on Pulse. If the flagship in July has real agents running real event-layer tasks alongside the organizer, the integration story between MentorMates and Pulse moves from speculative to demonstrated.
The question I want to hold myself to for the rest of March: every calendar day between now and July either feeds the Hong Kong flagship or gets pruned. Inbound is fine, but inbound has to serve the event or it moves to the post-July queue.
The check I'm making on myself: sponsor deck broken into three cuts by the next mastermind, venue confirmation by end of March, and the ambassador network pushed by me personally, not outsourced to a calendar.