Some observations from referring 63 founders to @alliance, maybe helpful for others:
1. Too many people building prediction markets still, what you think is a novel company is a weekend feature for incumbents who've raised a billion dollars and already have distribution.
2. Too many perp dexes and leveraged trading products still, their "novelty" is usually stacking leverage on already leveraged assets aka cascading catastrophe for users. If you burn the user once, they won't return.
3. Lot of agentic use cases, the low hanging fruit (agentic commerce, agentic coordination/unification) is less exciting personally. But a few novel applications across healthcare, science, gaming stood out to me.
4. Solana and Hyperliquid reign supreme, Base is a close 3rd. No Monad, MegaETH, Ethereum, etc. This is maybe an Alliance alignment issue or a larger preference trend for builders.
5. What many builders believe is unique and new is most likely already being done by another person in my DMs. The immediate opportunities are very obvious and $500K might not be enough to take on Stripe and Ramp, find new avenues where you have an unfair advantage and asymmetric opportunity.
6. Not a lot of "cypherpunk/privacy" builders, I think there's still massive opportunities in resilience, networking, shielding, identity for both crypto and AI use cases that I would've loved to see more of.
7. Manners. A please, thank you, and professionalism in cold messages go a long way.
Fun fact
One of the most interesting stats I recently came across was
An analysis of Solana apps after they went multichain
Almost every single time, opening to new chains did not drive meaningful revenue
The lesson is clear, build and double down on Solana
Post quantum secure simplex is running on invalidator cluster 1 with Falcon signatures.
Main changes are obviously to vote messages and then, for completeness (liveness under partial synchrony) what do we do when we didn't receive the votes for a certificate. Simplest thing to do is basically do what we do when we don't have the block: ask other validators for the missing info.