Founder and Partner @Standard_Cap, Partner Emeritus @ycombinator
SF Bay Area
Joined September 2007
- Tweets2.9K
- Following1K
- Followers356K
- Likes10.2K
Pinned Tweet
Applications for the latest Standard Capital Series A funding cycle are now open!
Apply by July 21, hear back by July 31.
Modern problems: trying to figure out if someone emailed you something that seems a bit strange or their agent did.
Dalton Caldwell retweeted
there's really no limit to what agent swarms are able to do if they share context. we tasked 128 agents with access to the same shared file system to self-organize and build their own version of factorio.
they broke up the problem, claimed tasks, and did their own testing.
because we used serverless tool calling, the agents were able to get access to exactly the amount of CPU they needed on a per-tool basis, spiking to over 100 cores simultaneously, but because they were only charged for what was used, the total amount of compute required was on the order of ~$0.50
archil implements leasing internally, which allows the agents to write to the same directories and files without stomping over each other.
Dalton Caldwell retweeted
the newest frontier is agent swarms. watch us launch 700 sandboxes -- all connected to the same file system -- in 10s, then have them collaborate and build a city on that file system
i used sandboxes here, but if i used serverless tool calling? the total cost of this compute would be $0.28
the future is packing agents 40x tighter on each CPU and giving them access to the same shared file system context
could it run like this but kind of on the 'same' file system? xitter.cf/AniC_dev/status/211120…
Dalton Caldwell retweeted
we're about to witness a rapid reshuffling of the "ai infrastructure" as people realize that the tools required to run a single agent are very different than the tools required to run thousands of agents that need to coordinate
the bedrock of what that future looks like? real, distributed file systems that can be mounted from many locations simultaneously in milliseconds.
for the past year, most agent data has just lived on the local disk of the sandbox (persistent or otherwise). this will not scale as agents work with larger data sets and coordinate more amongst each other, you need *shared storage* which allows multiple sandboxes to access the same data
luckily, we've been building for this future for years at Archil, with the worlds fastest shared file system (approaching local disk) that can also reduce the amount of vcpu required for agents by 40x with serverless tool calling
Do any of the sandbox providers give me the ability to:
- have N sandboxes running with disks that...
- can be mounted on another sandbox (at least 32 at a time)
- without knowing those mount points before the sandbox has started
- and being able to vary them throughout the lifecycle of that one sandbox?
im basically looking for a way to have a parent sandbox, N children sandboxes (where N varies, and has max X concurrency, which would be lower than max of N), and have those children write to a point on the disk that the parent could read from. N readers 1 writer semi-persistent disks.
Dalton Caldwell retweeted
Today, we're announcing Archil's new $10/mo plan for individual developers -- making it more accessible for builders to use the world's most advanced AI infrastructure.
To celebrate, we're hosting our inaugural [super fast] hackathon in a couple of weeks at Casco's offices in SOMA! We'll have $50K in prizes, and will be a great way to see how others on the frontier are building.
RSVP here luma.com/t9hsthtk and we'll see you soon!
🤖 Made with AI
Dalton Caldwell retweeted
My chat with @daltonc on Claude trying to kill our startup and the launch of MagicOS:
Dalton Caldwell retweeted
a convo about a founder's worst nightmare: dealing with competition from the frontier labs.
just stay focused. just keep winning.
p.s. we are hiring
xitter.cf/daltonc/status/2110777…
The question on a lot of founders minds: What do you do when Claude enters your market?
I sat down with @alexdanilowicz from @magicpatterns to provide some thoughts and advice for founders imagining how they would react and rise to the occasion.
In addition, we talk about Magic Patterns new MagicOS launch, a design agent built for the future.
The question on a lot of founders minds: What do you do when Claude enters your market?
I sat down with @alexdanilowicz from @magicpatterns to provide some thoughts and advice for founders imagining how they would react and rise to the occasion.
In addition, we talk about Magic Patterns new MagicOS launch, a design agent built for the future.
Dalton Caldwell retweeted
The consistent advice is, "don't hire big tech execs"
The fundamental problem is that they've been RL'ed by an environment in which success is measured by headcount. They will make your company bloated and ineffective because in their experience, that's what success looks like
Voice-first is an interesting approach, it's like pair programming with a world class designer (that also happens to be infinitely patient)
Today we're launching MagicOS. And we've come to the conclusion that designers are NOT getting replaced.
most of the industry moved to the same answer this year: give the agent its own canvas, let it run, hand you the output.
we made a different bet: the canvas should be yours and the agent is your collaborator. that’s MagicOS.
most AI design tools are a linear conversation: prompt, generate, iterate, repeat. a hundred prompts in, you might have a great final design, but the reasoning, rejected directions, and inspiration that got you there are buried in a chat log nobody will ever scroll back through.
MagicOS is a persistent, spatial canvas. Your PRD, inspiration images, design system, variations, rejected directions, final picks - all of it lives together, all in one place, instead of stacked on top of itself and lost.
it’s voice-first. talk to it in real time and watch work land directly on the canvas as you speak.
it’s also multiplayer. your whole team works on the same canvas together.
and it still runs on your design system and components because none of this matters if the output doesn’t look like your actual product.
three months ago, this was internal only at Magic Patterns, my co-founder's weekend project. it’s now live for everyone.
Design everything, everywhere, all at once.
Dalton Caldwell retweeted
"Great founders turn being uncomfortable into motivation, and it fuels them." - @daltonc
youtu.be/Bsxa_Yg1Xcw?is=hEz7…
Dalton Caldwell retweeted
When you talk to successful founders, they always wish they'd hired fewer top managers from outside. I don't think I've ever met a founder who wished he or she had promoted fewer people from within.
Dalton Caldwell retweeted
Thank you, Dalton! It was a pleasure getting to share what Jeevy's working on.
We recently led the @jeevyfab Series A, and so I recorded a video for the @Standard_Cap channel to talk about it.
Jeevy Fabrication builds the physical things that get rockets off the ground: tanks, piping, fluid systems, and structural assemblies. Their customers include SpaceX and Lockheed Martin.
Jeevesh previously worked on Starship ground support equipment at SpaceX and was previously at Rondo Energy, and they're now using AI and a data-rich coordination layer to fix the biggest bottleneck in hardware: getting parts on spec and on time.
Dalton Caldwell retweeted
Introducing AgentID by @agentmail, "Sign in with Google" for AI Agents
Millions of agents already use AgentMail. AgentID lets them sign in to your app.
Integrate with the prompt below, reply with a screenshot, and we'll give you 3 months free of any AgentMail plan.
We recently led the @jeevyfab Series A, and so I recorded a video for the @Standard_Cap channel to talk about it.
Jeevy Fabrication builds the physical things that get rockets off the ground: tanks, piping, fluid systems, and structural assemblies. Their customers include SpaceX and Lockheed Martin.
Jeevesh previously worked on Starship ground support equipment at SpaceX and was previously at Rondo Energy, and they're now using AI and a data-rich coordination layer to fix the biggest bottleneck in hardware: getting parts on spec and on time.
