
startups, growth, internet money founder @yumifinance ex. https://xitter.cf/t.co/seHPNfOIpz (1m$ peak ARR), @0xoogabooga (1bn$ in lifetime volume) 22yo
Hong Kong
Joined December 2023
- Tweets5.2K
- Following634
- Followers1K
- Likes6.1K
Pinned Tweet
LLMs began with a data warehouse humanity had already spent centuries producing - the internet. But robotics had almost 0 data. For a robot to learn, somebody has to create the action, instrument the body, record the environment, and label what happened
I swear to god if I see another "best ideas for a startup" on my TL im logging off twitter
algo has to punish ppl larping please
Wait until he sees crypto launched
I hate that launch videos have become larp-contests over who can buy the most QTs instead of who makes the best video.
desperate founders pay slop agencies $200k-$500k, who pocket half and spend the rest buying QTs.All this just to convince investors they've gone viral. It's a great business for the agencies who has no work, but absolutely waste of capital for founders who are sometimes investing like 10% of their total raise.
this scam is only possible on X as they don't have a minimum window of 3s to count the views. As long as it loads on your screen, wether or not you play it, views are counted. On top of that they aggregate all the views to the main post.
a great way to identify such posts is look at the views to engagement ratio, it will be way off. Like 5m views, 1k likes and 1k RTs, 32 replies. let's not even talk about conversions or adding MRR, as it doesn't exist.
if you think your company has PMF because you larped some QTs on X, you'd believe you're popular amongst women after visiting the red light area.
if you're a founder considering to launch. invest all your budget into video product, great writers and doing something newsworthy in your industry. don't be sheep.
The appearance of LLMs has led to re-pricing of the whole SaaS market but what about crypto?
Most crypto products are purely code + distribution (TVL), but as we saw with 99% of the cases - TVL/liquidity is not sticky (cue @berachain's TVL graph)
The only sticky liq is from heavily regulated LPs that can't put it anywhere else + require a lot of trust into the project. So old behemoths like @LidoFinance, etc. will stay
However LLMs have led to the complete repricing of the crypto memecoin market - having a working project in 21' or even 24-25' was already considered enough to signal "legitness" and get your coin to 1m mcap.
Now that's completely gone and trading is way more social now as that's the only "trust" signal when code is worth 0. Adjust accordingly.
I see 2 ways with the AI market moving forward
a) (the unlikely one) LLM will continue to improve gradually and will be more and more useful. We'll still see a huge correction in pricing and amount of financing AI receives but it won't be as drastic. A lot of companies will continue to benefit from AI and blue-collar workers (esp. for jobs that don't require many soft or creative skills) will be replaced
b) (the likely one) The architecture of LLMs is fundamentally unsuitable for deep knowledge work. Basically Yann LeCun's argument.
This makes sense bcuz
> We are already out of data
> More and more ppl are starting to realize LLMs can only do the easy 80% of the work, while the hard 20% still requires humans
> Tokens are becoming commoditized thanks to Chinese labs (I love them)
Then we won't see a step function improvement in LLMs anytime soon and it will likely take a considerable amount of time for us to find a new AI architecture that works. Most of the valuations were too optimistic - some companies will still survive but will be a choppy market. The re-pricing of SaaS will continue and we'll see (and are already seeing) more money being poured into fundamentally hard industries like healthcare, energy, space, etc.
Showed some tweets to a person who hasn't used X
His first response was:
"So basically everyone talks about how to launch products, instead of actually launching them? Larp era"
> I go on TikTok try learn America culture.
> see much young white America peoples say "six seven. hahahaha. six seven"
> I be confusion. "What this six seven"
> I ponder long time. I realize profound wisdom.
> America white young peoples be despair. They know they are fucked. Only chance for they have good life is if every thing collapse so they can rebuild.
> six seven really mean "six point seven percent interest rate"
> If Fed raise interest to 6.7%, economy collapse. Every thing begin new.
> phoenix can not rise from old brick building. Phoenix only rise from ASHES
> "Six Seven" really be young American peoples war cry for crash every thing.
oh my god. I too support:
"SIX SEVEN. SIX SEVEN. SIX SEVEN. SIX SEVEN"
StarCap wants to bring private space investments onchain.
I sat down with @Wajahat, co-founder of @Starcap_xyz.
He breaks down StarCap’s investment thesis across orbital data centers, space-based energy, and microgravity manufacturing. And why he believes falling launch costs could unlock a new generation of businesses.
Timestamps:
00:00 — Introduction: meet Waj, co-founder of StarCap
03:25 — StarCap’s origins: from tokenized equities to space investing
07:23 — Why StarCap is starting with a tokenized fund
09:56 — Are crypto investors ready for private-market exposure?
14:19 — Opening access to companies before they go public
19:56 — Why StarCap believes space is at an inflection point
22:21 — Three investment pillars: compute, energy, and manufacturing
23:05 — The case for orbital data centers
25:00 — Target companies, funding stages, and valuations
27:37 — How StarCap plans to access competitive investment rounds
28:39 — Financing space companies through onchain credit
29:28 — Supporting portfolio companies with research and media
32:51 — StarCap’s growth strategy and investor education
36:32 — Private fundraising and plans for a public phase
38:39 — StarCap’s roadmap: growing AUM and expanding its products
41:17 — The future of sector-specific tokenized funds
43:42 — Building StarCap’s brand for a crypto audience
45:47 — Space companies and other investments on Waj’s watchlist
51:06 — China’s space industry and StarCap’s global investment scope
53:14 — Navigating regulation and national security constraints
Today I went down the rabbit-hole of "AI employees"
There's at least 10 different ones: Viktor, Lindy, Mio, Jessie, James, Junior and more
Distribution is, indeed, the only moat
Jev (@typesafeai) is the future of love
i used Jev to automate finding baddies on Tinder
> specify your type
> jev decides if you should like, dislike or superlike
> gives u the best 1st message
(i still got 0 matches)
Not being cringe enough is the #1 thing that's keeping you broke












