Why I stopped going to startup events — and went back

For about eighteen months I stopped going to startup events. This was, in retrospect, the right call. They had become unwatchable. Every panel was three founders who had raised in 2021 explaining how they were "leaning into AI." Every networking hour was a bartender being pestered by people who wanted to "trade Calendly links." Every demo day was either a YC clone or a YC clone running on a smaller stage. I gave up.

Six months ago I started going again. Not all of them — most are still bad — but enough, in the right way, that I want to write down what changed and what I look for now. Mostly because I keep getting asked about this by founders who hit the same wall I did.

What broke

I think the honest version is that the 2021 cohort overran the events. Too much money, too many people who weren't building anything, too many panels organized by VC firms that needed to be seen organizing panels. The signal-to-noise was terrible.

Then through 2023 and 2024, the layoffs and the AI shift kicked in, and the events became weirdly defensive. Half the room was looking for jobs while pretending they weren't. The other half was AI-pivoting in real time. Nobody was talking about anything they actually believed because everyone was triangulating against the others. Networking conversations had a specific cadence: "what are you working on" — vague answer — "interesting, interesting, what about you" — vague answer — exit.

I went to one in late 2024 in Austin where I genuinely could not find anyone who could describe what their company did in one sentence without including the words "agentic" or "infrastructure." I left after twenty minutes and stopped going.

What's different now

Two things, both subtle.

First, the AI cohort has — partly — split into people actually shipping versus people still posing. The shippers are easier to spot now. They have specific customers, specific revenue, specific things that don't work. The pretenders mostly stopped going to events when the funding stopped flowing. The ratio is meaningfully better than 2024.

Second, the events that survived are mostly small. The 500-person summits at hotel ballrooms are largely dead or dying. The 30-person founder dinners, the 80-person demo nights at coworking spaces, the SaaStr-but-make-it-Austin breakouts — those are full and the conversations are real. There's a Mercor-related dinner here that's become a fixture, a couple of recurring AI infrastructure happy hours, and the usual roster of YC alumni meetups, all small, all word of mouth.

The defining trait of the events I now go to is: under 100 people, organized by an actual founder or operator (not a VC firm or an event company), and the invitation came from someone I trust.

What I look for

These are the heuristics. They're not perfect but they've cut my dud rate from maybe 80% to maybe 30%.

Size cap. Anything over 200 people is a tradeshow now. The conversations don't happen, you're standing in a corridor waving at acquaintances. Aim for 30 to 100.

Who's organising. A founder or operator running an event is doing it because they want to talk to other founders. A VC firm running an event is doing it for deal flow. An event company is doing it for ticket revenue. The first kind is meaningfully better, the second kind can be useful occasionally, the third is reliably useless.

Specific theme. "Founders dinner" is fine. "AI Infrastructure dinner" is better. The more specific, the more likely the people there have actually thought about the same problems you have. Generic "startup networking" gets you the bingo card of side hustlers and recruiters.

No sponsor logos on the invite. Not always a hard rule but a useful filter. The events that have to be sponsored to exist tend to optimize for the sponsor's interests, which is rarely yours.

Returning crowd. If you can't imagine going back to the same series next month, it's not worth the first time. The repeating series — first-Wednesday founder breakfasts, monthly AI dinners, the same group doing the same thing — those compound. One-off conferences mostly don't.

The Austin specifics

For what it's worth, the Austin events I go to now: the recurring AI dinners (you have to know someone), Capital Factory breakfast for early-stage stuff (about half useful, the size has crept up), Republic Square's founder happy hours, the Trans-Pecos crew's occasional gatherings. SXSW I went to and bounced off — the signal is gone in that crowd, it's mostly the marketing-conference cohort with a startup veneer.

If you're in London, NYC or SF, the same pattern plays out — small, specific, founder-organised. Rifio's startup events page for London has been useful for finding the smaller stuff that doesn't market itself, and the same pattern works in SF and NYC. The search picks up the small founder dinners that aren't on Lu.ma's front page.

What I'd tell my 2024 self

Stop going to the big ones. Stop going to anything sponsored by a Series-B fund that needs deal flow. Stop pretending that volume is the answer — going to four events a week at the wrong scale was making my work worse, not better, because I was performing all day instead of building.

Go to one or two small things a month. The 30-person dinner where you actually have a conversation with three people. The 80-person demo night where you talk to one founder for forty minutes. That's the unit. The rest is noise.

I'm not going to pretend events are the highest leverage thing a founder can do. They're not. Building is. But the right events at the right size with the right people will recieve you a couple of meaningful conversations a month, and over a year that compounds into hires, customers, advisors and friends. The wrong events get you nothing except a hangover and 47 LinkedIn requests from people whose names you can't place.

Choose accordingly. And, frankly, please stop going to anything where the first slide of the deck says "in today's rapidly evolving landscape."