
Gen-AI Dad. Building FounderOS, Signal. @runtechclub
Los Angeles, CA
Joined October 2008
- Tweets1.6K
- Following395
- Followers298
- Likes711
Pinned Tweet
one experiment, one check-in, one run at a time.
to challenge your hypothesis and show you where to focus.
founderos.run
If you’re building something massive, with venture scale, it must have the ability to dominate a huge market, not just exist inside a large TAM.
23
If you’re building AI for fill in the ___ (GTM, legal, media, production…), packaging your unique insight, process, skills. You are still the product. AI is the feature and multiplier. Your customers will still want to work with you, the human.
21
If you’re building something that connects everything into one place, email, calendar, Slack, messages, sheets… that’s a feature. Not a product.
28
Hard filters:
-Hype formats: kickoff parties, mega-mixers, summits with emojis, anything celebrity
-Vendor lead-gen dressed as community eg free coffee
-Wellness filler with no builder substance.
-pitch clinics and guru-talks.
1
1
22
Scoring criteria:
- Proof>pitch. show real work, through teardowns, or technical workshops.
- Diligence on logic. real operators or investing partners with POV. not panels of generalists.
- Trusted venture partners. credible funds.
- Room size. S to M, curated or app-gated.
1
1
20
Will drop 2 discord threads to help navigate during TechWeek and foster real connections via @runtechclub
Replying to @Techweek_
@Techweek_ is coming in hot. SF+LA
Here’s how not-to get lost.
Spoiler: don’t ask your agent to auto signup.
Instead set hard criteria and filters, then connect the TW MCP to curate and verify.
Sharing my framework below.
tech-week.com/mcp
2
2
43
Doesn’t matter what you’ve accomplished in the weeks, months or years before, it’s back to zero — every day.
We’re entitled to nothing. Keep learning. Keep growing. Keep stacking.
62
Clarity is rare. Conviction is hard earned.
The future of tech/venture will run on proof.
We are building that future.
— @runtechclub —
64
Watching this one closely.
Training custom models per customer is a much harder bet than instrumentation.
Signal — @raindrop_ai
Verdict: Track
What must be true next: custom per-customer models beat LLM-as-judge, in production, at scale.
linkedin.com/posts/dariusfon…
1
110
I get it. It’s easy to fall in love with your idea and work tirelessly to make it work. Then you read about lean startup and fail fast — only to find the fail part to be true.
linkedin.com/posts/dariusfon…
1
59
Frictionless comes with a cost.
We consume more and more. Remember less and less.
#AIDAD
28
Darius Fong 晧中 retweeted
Conviction is earned. Signal Run Friday. @Techweek_ #NYC |
VCs in the room: Wildwood Ventures | @PrimaryVC
partiful.com/e/OiFkcaWT26YBs…
Darius Fong 晧中 retweeted
Replying to @sekaiapp
@sekaiapp by @lucky_z2 announced $20M Series A led by @khoslaventures + @ConnectVentures Day 1 of @Techweek_ #NYC 🙌
The one number that truly matters: creator-to-consumer ratio. Session length measures consumption. Not creation.
Full DILA run → signal.day/cases/sekai










