startups, growth, internet money founder @yumifinance ex. miki.digital (1m$ peak ARR), @0xoogabooga (1bn$ in lifetime volume) 22yo

Hong Kong
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
Article

How to train an LLM vs How to train a Robot

TLRD; it’s the same process; Train an LLM: Get data → learn → deploy → collect better data → repeat. Train a robot: Get data → learn → deploy → collect better data → repeat. This article will

5
4
30
3,099
Am I the only one that's getting "This model is out of capacity" like non-stop on gpt 6.1 sol
6
166
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"
6
158
Your net worth in the age of AI is the number of friends you have
2
113
In urgent need of a good coworking place in Shenzhen Bonus point if it has AI/Physical AI builders
1
1
4
191
Mikhail (adhd arc) retweeted
> 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"
26
33
531
48,779
Mikhail (adhd arc) retweeted
Community note
The video is an animation by artist R.J. (@RJ16848519) from his Screen Time series, shared here without credit. x.com/rektober/statu… x.com/connoissurreal… x.com/kikanicolela/s…
369
23,090
161,220
5,063,327
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
6
6
29
3,342
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
1
7
245
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)
1
3
184
Not being cringe enough is the #1 thing that's keeping you broke
3
120
I used Claude to find 21 leads ALREADY complaining about my competitors. For less than 10$. Here's how: > Identify your competitors > Use exa.ai + apify (or simply @MonidHQ) to scrape Trustpilor + G2 + ProductHung + Reddit + X + LinkedIn. > Jev (other models work just as well) to judge if a review is a complaint or not > Resolve the humans: Name → company → email via Apollo + Hunter + ContactOut + LinkedIn. > Enrich every company and score by your ICP Those are some of the warmest leads you can find, like a Head of Product at a 180-person company who's disappointed enough with your competitor to publicly call it out If you want the full prompt you can paste into Claude to do this all for your - comment and I'll DM you!
5
217
Deel just shipped its own harness. $140M ARR in 90 days. zero new hires. 8k agents doing the work of ~600 people. It's also one of the few multiplayer AIs (said it's the future a while ago) Let's see how it stacks up to the rest of the market!
EXCITED TO LAUNCH: Akai (akai.run) Deel added >$140M ARR in 90 days without increasing headcount by automating~600 Full Time Employees' equivalent in work with Akai. Akai was an internal tool to automate our painfully repetitive operations in Finance, HR, Accounts Payable, and Compliance, etc. We never intended to make this a product. But we watched revenue per employee grow from $130K to $215K We built >8k agents that do the work of ~600 employees It had such a dramatic impact on our business that today we are launching it for everyone. How it works: Say you're automating payment reconciliation: 1. Record your screen while manually matching a messy transaction and Akai will capture your screen, voice, server requests 2. Akai will see that you pulled unformatted wire transfer info from an archaic bank portal, put it in some excel sheet, checked NetSuite invoices, payment history, and put a ticket on Zendesk 3. Akai reads between the lines and build a workflow + steps + conditional guardrails. It learns tacit edge cases, like resolving malformed invoice references without you writing a single regex 4. Simply connect NetSuite, your ledger, Zendesk, PSPs, and even legacy bank portals with zero API access 5. Run the workflow and tell it what to adjust in plain English: "strip slashes on wire memos and auto-apply partial payments." It adapts instantly 6. Once it works for you, add 100s of colleagues. Your entire payment ops team forks and extends the workflow for new PSPs, secondary ledgers, or regional settlement rules 7. We automated 85% of our payment reconciliation end to end, eliminating 500+ hours of soul-crushing manual grunt work every single week. Claude Code/Codex can't do this in multiplayer mode. Every person rebuilds the same skill from scratch in their own way. Deel built Akai to: 1. Understand backend operations edge cases (it had to work for our 7000 person team first) 2. Collaborative across 1000s of employees 3. Self-Learning from millions of runs 4. Optimises cost and gets cheaper every run We're so confident that we're announcing an Automation Guarantee: If our engineers can't automate a thousand of hours of work in your first 30 days, you get a full refund. Book a demo: akai.run if you're an exec at a company with hundreds of employees
5
312
Thought virality on X was organic? Fuck no. Half the “overnight” AI launches you saw this year (@higgsfield, @FactoryAI, Wonderful, Parloa, @typesafeai) ran through paid creator networks that seed replies and quotes in a 4-hour window. How do I know? The agency behind them literally said it on their website Welcome to the pay-to-win era
2
7
264
grokbot is like a college grad-level coworker except at times it's autistic and tends to forget everything you say i still haven't found a decent way to orchestrate agents and I think it's a fundamental LLM problem that can't be solved by a better wrapper (Hermes, OpenClaw, etc.) imo we'll soon see a lab that makes agents actually useful for prod and it might be a totally new player
2
6
206
Jev + Tinder I think I officially found the best usecase
1
7
643
The biggest limitation of grokbot is that there's only 1 PC for all the agents You can't do stuff in parallel at all
1
4
105
keyword lead gen is dead ~15 hard-pain leads from one scrape of competitor engagers they were already arguing about pricing/support just had to shoot them a dm P.S. i'll write a longer post about the mechanics of this (basically Jev + @MonidHQ + @browser_use to scrape G2, X and Reddit)
1
5
217
This is cool, but how big does your TAM need to be for you to actually benefit from the speed? If ChatGPT/Claude spend 5 min analyzing 40 leads/accounts, they can analyze 800 in an hour and 18400 in a day - way more than u can cold email effectively When everyone has access to speed, maybe to get an edge you need to slow down and think what to write for 5 good cold emails, instead of blasting 18k
JEV is a cheat code for outbound. Deep lead qualification and outreach prioritization used to cost us a fortune. GPT and Claude analyzed prospects one by one. It took minutes. JEV just analyzed 40 high-intent prospects in 5 seconds. Total cost: $0.01. Now live on @GojiberryAI + MCP.
4
287
This is why I'll keep saying that hallucinations are a trillion-dollar problem in AI and the biggest bottleneck to production use
instinct made up my middle name, and now I can’t board the flight
4
278