If the AI miscalculates, and building crashes and people die - who is held responsible?

The plans are the input here, not the output. The AI prepares cost estimates. It isn’t responsible for what is getting built.

So basically you built an estimation wrapper ready to use - without any guarantees whatsoever? Someones still needs to double-check?

why is that different than other LLMs

Estimation is by definitely not final. Did you watch the demo? It’s a workflow tool.

You can trivialize anything like this, salesforce is just a Postgres wrapper, doesn’t mean it’s not valuable.

Typescript with strict types and ultra-tight eslint config can give Rust a run for its money.

So instead of saying "We removed fsync" you should say: "We redesigned the database write path to avoid paying the full fsync durability cost on every write"

Can't remember when I last heard TOCTOU term being used. Nice :)

I also asked what happens when a power loss happens.

So basically, you are writing data without guarantees it's actually written? "YOLO mode" but for data written to a device?

Would you be so kind to explain what happens in a power-loss scenario?

Why does everyone feel the need to reinvent the wheel every few days?

Too little too late. I've already fully migrated to Fedora 43.

Thank you for this detailed writeup. If it's not a secret - what is your monthly ballpark figure. I'm planning to deploy an instance during the weekend, for light - medium usage. Can't gauge costs.

Can I tag along to the main question and ask - what are some good OpenClaw guides or plugin/skill collections that you think are must-use?

So much AI generated bullshit online, it's really hard to find good guides from bad ones.