I am creating ai website builder https://playcode.io/ai-website-builder, the full stack part.

I have been working on it for 10 years already.

Yesterday I faced 5h window limit for the first time. I was surprised. Max 20x plan. Usually I work 12-15 hours per day 7 days a week with no limits. But yesterday it was under 3 hours… what a pity.

Well, 70% of my commits are “123”.

Congrats guys! Would share some technical details, I bet you have great stories to tell. Let’s, what is forking? You completely copy disk, make ram snapshot and run it? If CoW, but ram? You mentioned 8GB ram vms. Sounds like impossible to copy 8Gb under 500ms, also disk?

So fork time is actually O(1) with VM size, its 500ms even for 64gb + disk. We're using some pretty weird COW techniques to pull it off.

O(1) What! What might bring it down to say 10's of ms? Looks like its some kind of optimizable wall that its 500 for everything.

Like with 10ms then online replication/backup — analogus to litestream for sqlite — but for in memory processes becomes feasible, no?

We're actually median under 500ms — ~320ms median — I just didn't want to piss of hacker news with over estimatation.

We have another set of optimizations that we believe can take us to ~200ms in the next few months but beyond that we're pretty much completely stuck.

Realistically other sandboxes will be able to get there before us because we've chosen to support so much of Linux/if you don't run an operating system or don't support custom snapshots that is much easier.

Insane. Does it possible to fork to another bare metal machine? Maybe multi region as fly io. If not, I bet you have huge disk sizes on your machines to store all the snapshots (you said, you store them and bill only for disk space).

So forking across multiple nodes in that speed is not possible — we run extremely beefy nodes in order to avoid moving VMs across nodes as much as possible.

We are researching systems of hot moving VMs across VMs but it would have very different performance characteristics.

Yeah, I see. Is it possible to get a corrupted state? Let’s say we had realtime database actively writing at that moment?

It is impossible.

Our tech is not decades old so there is a chance we've missed something but our layer management is atomic so I'd be shocked if you'd be able to corrupt state across forks/snapshots.

Also modal.com, I saw a few more as well.

I use it ultra extensively and it works absolutely fantastic. Sometimes I think: "people are right, it is worse now" and then realize it is mistake, poor context or poor prompt. Garbage in, garbage out. No, it works not worse, but better.

I built entire AI website builder https://playcode.io using it, alone. 700K LOKs total. It also uses Opus. So believe me, I know how it works. Trick is simple: never ever expect it finds necessary files. Always provide yourself. Always.

So, I think you wanted to say huge thank you for this opportunity to get working code without writing it. Insane times, insane.

Huge thanks for 1M context window included to Max subscription.

> No, it works not worse, but better.

"Is it me who is wrong? No, it's everyone else!"

700k lines of code - something to brag about?

I was laughing so much. Thank you. Unexpected tech!

Are they affiliated with RAM sellers?

What do you sell? Names? Names for what? Site is kinda laconic…

MCPaaS serves persistent AI context via the Model Context Protocol. A namepoint (mcpaas.live/yourhandle) gives your AI instant project context — no re-explaining every session. Works with Claude, Gemini, Cursor, any MCP client.

Claim yours, free or paid: https://mcpaas.live/claim

The globe shows where the edge binary executes. More at https://mcpaas.live/about