arth retweeted
"Did you know you can reduce tokens used by 6% by jumping through a bunch of hoops, making your life worse and also making your agents dumber? Why isn't everyone doing this!"
- Half of current "startups"
arth retweeted
we're also hiring designers and design engineers, but never have i ever given a test or take home task.
you can tell the difference the way they talk about their previous projects.
i must say it takes skill to identify greatness in others too.
arth retweeted
Build for the 5x
dx - developer experience
ax - agent experience
bx - bot experience
px - physical experience
ix - influencer experience
My (long) take on lovables/replits.
I meet lovable+supabase customers almost weekly now. They contact me and I go get a coffee or do a call with them (many of them local) to ask how is it going. What they're building. Where the pain points are, etc.
Most of them have no technical background. I check their projects for vulnerabilities and so far they've all been safe. Lovable catches most issues early. I have talked to 10+ different customers in the past few months so my sample is not very big.
What's interesting: most of them have customers already. This is because instead of developer brain they come from sales/marketing/generalists and they built something where they had existing demand. Something that either complemented their existing business or they had an upper hand in gtm. They validate early and talk to customers a lot.
They don't care about token cost optimization. Getting most tokens out of $ spent. This is a developer brain thing. They only care about $ spent vs $ generated and getting stuff done.
This is a new type of customer that didn't exist 2 years ago. Seems like a big market, or at least we haven't seen the ceiling yet. Most of the narrative going on here is completely unaware of this customer and think that lovable are selling to regular devs. At least in my X bubble.
It seems to be all SMB. If you know anything about SMBs you know that what kills them is usually bad hiring or bad GTM. They don't get to raise millions to subsidize bad hiring or bad gtm. They find that with lovable they can finally fix that problem.
I own part of a small family catering business. I've built a ton of tools that help us save money and have more time to work on what's important. Not with lovable, but I can 100% understand other SMBs that find that they have access to the same solutions now.
Seems to me like it's incredibly early and there's a ton of opportunity to build for these customers. But they're not on X.
arth retweeted
So many devs seem to think that the $100 and $200 subscription plans are how Anthropic and OpenAI make their revenue.
There are companies spending over $500 MILLION a year with Anthropic. Your subs aren't even a drop in the bucket.
arth retweeted
Meet Pancake, the Vercel for GTM.
Try it at pancake.ai
Your agent ships your product. Now it ships your go-to-market.
How? A GTM brain that knows your business, 50+ data providers to find the right customers and tools to reach them.
Your next customers are one prompt away.
(did I really have to jump off a plane to say that? questionable, but I survived and now you're reading my tweet)
arth retweeted
AI will make everything free, including itself. The last domino to fall will be free energy, which is humanity’s final frontier.
arth retweeted
DHH is fundamentally right about Rust. For context, Vercel has been undergoing a Rust-ification (carcinization, technically 🦀) for a while.
One of the first projects we migrated was Turborepo, from Go to Rust¹. The migration completed, but the RoI was actually quite controversial internally.
While Rust was in our eyes better for low-level OS access, something crucial for a build system like Turbo, the human migration costs were very sustantive.
Go is very fast. It's beautifully designed. It's easy to iterate on. We were very conflicted about the migration, because it was *humans* writing the code, *even if we knew Rust was a better choice*.
The calculus has now changed. What's "best for humans" is no longer necessarily "best for business".
FWIW, it's also quite unlikely that Rust is the end-all-be-all toolchain. I'm quite certain there's greener pasture ahead, because Rust itself was designed before the 'supersonic tsunami' of agents hit.
¹ https://vercel.com/blog/how-turborepo-is-porting-from-go-to-rust
arth retweeted
i feel like higgsfield is somewhat of a scam now that i'm generating ai videos myself.
they're preying on young people who want to make a quick buck by engagement hacking the timeline. it's like the venture backed company version of being a course bro at scale.
the ceiling of what's possible using their ui is incredibly low. i don't trust anyone who says that they're actually generating videos using it.
the only practical way to generate at scale is with an api endpoint & coding agents. i say this from lived experience.
that said, the promise of the technology is incredible & they're probably going to make a really great api & capture a ton of value from ai video. i just wish they did less gimmicky marketing & more user education.
but alas that's not how capitalism works...
arth retweeted
Foldkit 0.162.0 and 0.163.0 are out!
- Swipe-to-dismiss Toast
- Explicit canonical URLs
- SSR builds no longer ship an unrendered HTML template
- evo → modifyFields
- keyboardShortcuts → keyBindings
foldkit.dev/blog/foldkit-0-1…
arth retweeted
Translation: "I'm about to IPO, so let's slow all other labs down so they don't announce a new model that slaps my valuation in half. "
Anthropic prob should withdraw their S1 filing to show they don't have ulterior motives
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
arth retweeted
companies building their own agent rather than exposing an mcp / cli for their uses is like if you had to download a separate browser for every website you use



















