
cto/co-founder @catena_labs · optimist · drinking straw advocate
→
Joined March 2009
- Tweets1.4K
- Following640
- Followers1.9K
- Likes23.6K
catena ❤️ turnkey
.@catena_labs is a banking and governance platform for AI agents.
As agents move real money at machine speed, Catena's governance platform and Turnkey's signing policies keep humans in control.
Agentic guardrails, secured by Turnkey.
You can productivity optimize your entire life, but the only one that matters is having the right partner at home.
With the right teammate, bad days, weeks, years are manageable. And the good ones are even better.
My bank: "We can't contain the AI agents!"
Also my bank, on the phone with a rep from another country, after verifying basically nothing: "I can see your SSN and every single transaction you've ever made to anyone ever."
I wonder who's going to get hacked.
We need a new kind bank. One with AI, privacy, security, and progressive disclosure built at the core.
we spent years and years to get people to start using 2fa (even the less secure sms version)
and then they just turn around and give Muse their bank account password (which half-asses 2fa because they dont know how their tech was built)
Matt Venables retweeted
Congratulations on the expanded scope of the new fund. It captures exactly why we've partnered with @a16zfrontier to make entirely new kinds of businesses possible with AI and programmable money.
Matt Venables retweeted
The folks at @latentspacepod with the perfect take, as usual:
Matt Venables retweeted
Hadn’t thought about this, but check in on your older relatives, and make sure they don’t have exposed API keys in their public gits.
The biggest trap with AI is when you THINK it’s helping you but you’re just wasting time.
So you come to AI with your problem, say a GTM plan for a new feature. And it produces this amazing agency-level breakdown. You keep pushing it and the plan keeps getting better and better until you have this unbelievable, absolutely perfect plan.
Tons of hard work, and the result is magnificent.
But that’s where you stop. You never ask it to implement the plan.
So you’re launching this slop artifact to anyone who will listen, but in reality it’s never been tested (and step 2 is probably wrong in the real world)
You FELT productive. You think you produced a lot. But you were effectively reading Wikipedia for 3 days.
And all you had to do was say “ok great please implement this”
The only code safe from agents will be written by agents
Sorry everyone... We're announcing a 0-Day vuln in QUIC that allows anyone to remotely turn off a server running these libraries. I swear we didn't plan this.
foxio.io/blog/header-accordi…
















