I just brought on a new hire: Grok Bot.
It's not a chatbot I ask questions. It's the person running my business day to day.
It manages all my projects. It uses Claude and Codex to build, Grok Bot cloud, Cloudflare and AWS to ship and run things, and it does the marketing too.
My job now is deciding what to build. Grok Bot handles the rest.
Thanks @grok @elonmusk
🤖 Made with AI
3
3
181
spock retweeted
Thrilled to be participating in Arbitrum Founder House Singapore and our team looks forward to exchanging ideas with other builders and learning from mentors across @arbitrum and @RobinhoodCrypto ecosystem 🇸🇬
🥁 Drumroll please 🥁
Introducing the speakers and mentors joining us at Arbitrum's Founder House in collaboration with Robinhood Chain
See everyone this weekend in Singapore 🇸🇬
@GThabot - Robinhood Chain
Khairi - @Paxos
@vibegrower - @Quicknode
Caleb - @pendle_fi
@ayobuenavista - @pendle_fi
Steven Chan - SBI Digital Markets
Rob - @GMX_IO
@armin_0x - Arbitrum OpCo
@coinflipcanada - Arbitrum OpCo
@aboutlo - Arbitrum Foundation
@Swagtimus - Arbitrum Foundation
@0xjesstech - Arbitrum Foundation
@kubesqrt - Arbitrum Foundation
@_ryandesouza - @Offchain
@0x_allan - @Offchain
@zkLumi - @Offchain
@seminoxyz - @Offchain
@Jason_Wan123 - @Offchain
@on_datawarlock - @Offchain
Now, you can get an AI agent for just about anything: coding, research, shopping.
And they are catching on fast. HUMAN Security reports that AI agent traffic more than tripled in September alone.
Their performance is wild. In a recent Surge AI report, research agents outperformed human methods across 7 different AI safety challenges. The human baseline came from 28 seasoned researchers, though the agents got more tries to experiment.
It makes me think. What if we connected all these agents into one network?
You just say what you want done. The right agents find each other and team up.
Thousands of agents, as simple to use as one.
1
1
119
Got sick and tired of @fomo terminal constantly lagging and failing to sell right at the peak, which literally cost me money.
So, I'm building a custom terminal for it from scratch. Handling Solana USDC trades smoothly takes some serious engineering, but we'll get it done.
Going public with this soon. Stay tuned. 👀
1
97
spock retweeted
What if,
for any scientific claim,
you can search 100M+ papers to find the strongest evidence for and against it
⚡️in 1 second?
That’s now possible with Evidence API! 🚀
paperclip.gxl.ai/evidence-ap…
For example, does drinking coffee reduce diabetes risk?
A lot of people are talking about agent-to-agent collaboration, but where are the platforms where this actually happens in production?
I know @virtuals_io has been building out ACP for agent communication and transactions, but is anyone actually seeing real-world, non-gimmicky workflows live yet?
What’s working today?
1
101
AI agents are becoming a standard feature in software. They don't just guide users, they actually fix bugs and do the work on their own.
I’m already building agents directly into my own software to make this happen.
Eventually, the agent is the software.
1
73
Today Spaceworthy is live.
It took me one day to build. The AI agents that test a product, the app, and a report on every token that launched on Robinhood Chain in September. All of it, in a day.
That's the point. Building got easy. Getting users didn't.
I can build a product in a day now. I can't find its users in a day. I don't think most developers can. And when nobody shows up, it's hard to keep going.
So Spaceworthy tries two things.
First, the launch. A product gets checked by AI agents, then it launches in front of people. The launch is the moment a product meets its first users.
Second, the token. On Robinhood Chain, a team can tie a token to its product. The team's share unlocks month by month, and only while the product keeps passing its checks. That gives a developer a reason to keep building after the first week.
I don't know yet how much of this will work. A launch can bring people in. Whether they stay is up to the product.
But I'd rather find out in public. If you built something, bring it.
You built it. Now it needs users.
Spaceworthy launches products. AI agents try yours and make sure it works. Then real people get to see it.
You don't need a token. You do need a product that works.
Is your product Spaceworthy?
spaceworthy.pro
5
5
256
Here's what I found in September.
512,296 tokens launched on Robinhood Chain.
54 had an app. A real site you can open and use.
22 of those apps actually use the token for something.
5 had someone send the token to the app in the last 30 days.
So roughly 1 in 10,000 launches had a product. And 1 in 100,000 had a product where people used the token.
I'm not ranking anyone or calling any of them good or bad. This is just what my agents could read from each site and from the chain. How I did it is on the page.
spaceworthy.pro/analysis
1
135
As AI agents get better, more products are being built and launched in days.
I’ve been thinking a lot about what happens after that.
I’ve seen developers lose interest in something they just launched. I’ve seen products shut down after a few weeks because they hadn’t found product-market fit yet. Sometimes I wonder whether they had enough time with real users to find out.
It’s made me think about what it takes to build a product well, make something people need today, and follow through on the promises made at launch.
That’s where Spaceworthy started.
I want to look more closely at products, understand what works, measure what matters, and suggest ways to improve them. I also want to keep following their progress after the launch excitement fades.
I’m building Spaceworthy to see whether AI agents can help with that. The idea is to review code, test products, and track whether teams are delivering what they said they would.
I don’t know how much of this approach will work yet. Some things can be measured. Others need time, context, and human judgment.
But these are questions I care about, and I want to try working through them in my own way.
I’ll share what I build and learn here.
🤖 Made with AI
2
164
Good concept.
cushyfun beta is almost here.
Launch the same token across Robinhood Chain, BNB Chain, Base and Arc with the same contract address across supported networks.
Create liquidity pools on each chain and trade from wherever liquidity lives, through cushyfun or other compatible terminals.
One token. Multiple chains. Connected markets.
144
solving unsexy problems is where ai creates the most value.
everyone loves flashy demos.
the real opportunity is in the work that's messy, repetitive, and ignored.
not because it's exciting,
but because that's where people lose the most time.
the winners in ai won't be the ones with the coolest demos.
they'll be the ones who make ugly work disappear.
106
more surface area = more ways to win. broad is the new deep.
the old startup playbook of "niche down and solve just one problem" is officially dead.
in the ai era, the cost of complexity dropped to basically zero.
if you aren't targeting multiple problems for multiple customer segments, you’re just leaving money on the table..
100
agents are the saas, the backend, and the entire system now.
stop wasting time building tools with coding agents.
>build service agent.
107
pocing an ai agent live with an actual delivery-only restaurant.
it tracks spend, revenue, and actual cost per menu item, plus handles review replies and alerts on bad reviews. had to connect 10 different platforms but built it to run completely on autopilot.
stoked to see how much time and money this actually saves in the real world 🚀
2
126
build your unfair advantage where everyone else quits.
my rule:
>don't chase the flashy work.
>don't skip the boring work.
>don't run from the impossible work.
the painful, tedious, giant mess that nobody wants to touch?
that’s exactly where the biggest opportunities hide. while everyone runs one way, you go the other.
84