Archil

Just got a look at what the @archildata team has been cooking 🔭 📦 Cold install (no cache): archil+pnpm finishes in 8.43s — local pnpm takes 20.6s. That's a 2.4x speedup just from using archil. 💾 Warm install (archil cache): once archil has cached your packages, install time drops to near 0ms. Local npm still spends 6s just installing. Faster installs. faster CI. How long before this reshapes sandboxes?

The future of AI is going to be built on sandboxes and file systems. We're thrilled to sponsor the @daytonaio compute conference next week, and chat with the developers working on the frontier of agents. Swing by and tell us what you want to see out of the next-gen file system!

See more on our docs here: docs.archil.com/sdks/typescript

Starting today, you can create, manage, and mount Archil disks directly from our Javascript SDK -- making it easier than ever to spin up infinite file systems for agents and projects.

See more here: docs.archil.com/reference/terr…

Archil now has a native Terraform provider to support creating and managing disks from your existing Terraform code.

file systems are the most important state management infrastructure for the AI era. but, until @archildata, there hasn't been a company focused on making the file system a first-class dev primitive. what features do you want from your file system? we're already tracking: - create a file system in milliseconds [ga] - skills that allow agents to get shared file systems without a human-in-the-loop - attach to multiple agents, running in any cloud [ga] - support running locally so you can debug the file system on your laptop [preview] - branching and forking - run any application: databases (SQLite, PostgreSQL, MySQL), git, or bash performantly so agents can choose the right tool for the job [ga] - ability to inspect the file system from the web [ga] - ability to synchronize the file system from S3 [ga] - attach the file system to places where you aren't running Linux, like just-bash [ga] are we missing anything?

alibaba just released "zvec", the SQLite for vector databases -- super simple, in-process, and supports thousands of queries a second. this means you can now run an entire vector database directly on top of any S3-compatible storage provider with @archildata. check it out! zvec.org/en/

archil now supports connecting s3 buckets directly to just-bash for read/write support anywhere that you have access to typescript. agents running on macs, Fargate, or AWS Lambda can now use S3 as an inifite, high-performance file system

In a full circle moment from my time at AWS, I got a chance to sit down with @QuinnyPig and talk through the problems that @archildata is solving for customers. We get to talk in detail about what this means for AWS customers, and (if you need a reprieve) I promise that we hardly even talk about AI. Check it out here: youtube.com/watch?v=rMcZ0K…

Why are file systems becoming the default context layer for AI applications, and are they here to stay? Read more on our blog: archil.com/post/why-file-…

the hardest part of using @archildata is needing use AWS. today, we're eliminating that requirement and introducing Archil managed buckets. now, you use Archil without any S3 configuration. if you also take advantage of Archil shells, you don't need any infrastructure at all.

s3, gcs, and r2 work great until you start getting 503 SlowDown responses. why do you need to design against these low limits? teams using @archildata don't have to. our high-performance storage layer supports tens of thousands of operations per second to even single objects.

hacker news doesn't think that SQLite can be run directly on S3. they're wrong. @archildata makes it possible to run any program on S3, without extra persistent storage, including complex read/write database use cases

really enjoyed getting to sit down with Gabriel and talk about @archildata’s journey and technology! give it a listen

wouldn't it be great if you could just go to the s3 console and run unix tools on what's in your bucket? now you can. today, @archildata is launching Shells, a one-click, ephemeral Linux machine attached to your S3 bucket so you can run vim, less, and grep right on your data.

Today, I’m thrilled to announce that @archildata is publicly available in preview. Infinite, shareable disks that are instantly connected to your data and radically simplify your system architecture. You can get access to the same high-quality data infrastructure that the frontier labs use with a single click right now.

Shout out to @ianlivingstone and @tnachen for hosting me on the Infra Pod last month. It was a great conversation about the inner workings of @archildata and where we see the data industry going in the next 10 years. Link to listen below 👇

we at @archildata have some exciting launches coming up in the next few weeks. if you're interested in using S3 / GCS like a local disk early, DM me for a sneak peek. 👀 everything will only accelerate from here

just found out that since this, i've become a top 50 user of Devin globally, now pushing ~60 PRs a day. AMA