Building apps on your laptop with Elements and Claude Code is insanely great. I'm still partial to the terminal, but watching Claude drive the hot reloading browser app in real time is magical. Elements is what makes the build loop extremely tight.

Hot deploy takes about 1s. Elements on your laptop speaks a p2p protocol to Elements on your boxes, over standard SSH. Deploy includes: upload, type check, compile, migrate db, tests, atomic release. Zero downtime, connected browsers hot reloaded.

40s later: the site is live with real time chat, auth and live data, across two load balanced machines on HTTPS with shared Postgres.

Paste your domain name, ip addresses and db connection info into Claude and it'll use Elements to do the deploy to two cheap Linux boxes and hosted Postgres. Initial deploy takes ~40s.

A few seconds later, we have an app, working locally inside the Claude Code desktop app. We'll click around and make sure it all looks good before we deploy to our load balanced cluster.

Next up, grab the demo app prompt for Replyhaven: a software company help desk. A live queue with real time conversations and chat, email notifications, auth and roles. We'll use Claude to build the app with Elements from an existing scaffold.

First, let's set up @digitalocean - 2 machine Linux cluster and 1 managed Postgres. A single small droplet is fine for small apps, but we're going beefy for this one: two load balanced machines with shared managed Postgres. Boom.

Here's Claude Code with a build client beside it so you can watch the state change. Claude fixes its own errors, with corrections from Elements in real time.

I originally built Elements for humans. But it turns out the same architecture works great for agents, if you make the data a bit more structured. Here's an example of what an agent sees.

Elements runs as an always-on project server. Your editor connects to it via LSP. The build command connects to the same server, so build results come back in microseconds. One server, less memory, less cpu.