Tomas Pastore

Building

[Tomas Pastore @tomasgpastore]

Are you still running hella agents on a MacBook?

Your laptop's glitching and lagged. Killed one to make room. You lost 2 hours of work. Then you ssh'd into a cloud box and explained the whole task again from scratch.

Ran into this problem myself. https://t.co/5UKugzjP5C fixes this, it's a native mac terminal that moves your claude code / codex session to the cloud mid-run. No re-clone. No re-auth. Same agent, same scrollback, same files.

Never start over
https://t.co/FxEoHa76yn

[Peter Steinberger 🦞 @steipete]

@hraness @DevinAI @zeddotdev I did this last year, that's not the future. Web's way nicer.

[Tomas Pastore @tomasgpastore]

@steipete @hraness @DevinAI @zeddotdev that is why i am building https://t.co/5emI4Szn2Y so anyone can run their agents in the cloud

[Tomas Pastore @tomasgpastore]

@hraness @DevinAI @zeddotdev When I tried to do that my computer got fried. I sign up for https://t.co/5emI4Szn2Y where you can run all your agents in the cloud, super fast.

[Tomas Pastore @tomasgpastore]

@CodeWithKolin We are launching soon, make sure to join the waitlist so you don't miss it. Best part, Murmello works with all of your agents, so you can bring your Cursor CLI account.

[Tomas Pastore @tomasgpastore]

I kept leaving my laptop open overnight so an agent could finish.

So I built Murmello: a Mac terminal that opens Claude Code or Codex on its own cloud machine. Start it, close the lid, pick it up later right where it was.

Get early access → https://t.co/CwKdFA7Bmr

[Tomas Pastore @tomasgpastore]

@opencode cheaper + faster models = way more agents. the next challenge is going to be computing limitation. https://t.co/04dnkx4Ydc

[Tomas Pastore @tomasgpastore]

@sama cheaper + faster models = way more agents. have yall moved your session to the cloud yet? https://t.co/CwKdFA7Bmr

[Tomas Pastore @tomasgpastore]

I’m a founder of a team of 1 person and 12 agents running simultaneously.

While building https://t.co/5UKugzjP5C, I set up the agents to communicate with each other, hand work off, and even have one agent act as a “manager” to keep time, check progress, and coordinate the others.

I wasn’t really managing 12 agents. I was managing one agent that was managing 11 others.

Running all 12 locally also completely maxed out my Mac’s memory, which is what pushed us to move the whole setup to the cloud (shameless plug, but that’s exactly what Murmello is built around: letting you move your AI work to the cloud when your laptop gives up without having to start over)

It’s been a fun experiment in what happens when you treat agents less like chatbots and more like an actual team you can delegate to.

[Tomas Pastore @tomasgpastore]

@cursor_ai cheaper + faster models = lots more agents. have anyone moved their session to the cloud yet? https://t.co/UaZJRRX4qK

[Tomas Pastore @tomasgpastore]

@cursor_ai https://t.co/UaZJRRWwBc

[Tomas Pastore @tomasgpastore]

@cursor_ai cheaper + faster models = way more agents. have anyone moved your session to the cloud yet?

[Tomas Pastore @tomasgpastore]

@ClaudeDevs the longer you let the agent run, the less you want it tied to your laptop this is basically why we built Murmello https://t.co/pHJbD9wm0E

[Tomas Pastore @tomasgpastore]

@OpenAIDevs cheaper + faster models means people are going to run way more agents

now we just need our laptops to keep up. try Murmello: https://t.co/FUDXY9VtyW

[Tomas Pastore @tomasgpastore]

@OpenAI cheaper + faster models means people are going to run way more agents

now we just need our laptops to keep up lol

built murmello exactly for this: try https://t.co/ByK0NBJMu9

[Tomas Pastore @tomasgpastore]

@ChatGPT Kinda wild that we have powerful AI models that can code for hours but we're still tying them to the battery life and RAM of one laptop.

Built Murmello to let you move the live session to the cloud instead: https://t.co/Pv806o0Yak

[Tomas Pastore @tomasgpastore]

@claudeai This model is getting crazy good. Now we need a set up that can survive having it run for 8 hours straight.

Exactly why I built Murmello for this: https://t.co/CwKdFA89bZ

[Tomas Pastore @tomasgpastore]

@AnthropicAI This model is getting scary good. Now we need a set up that can survive having it run for 8 hours straight.

Built Murmello for this: https://t.co/CwKdFA89bZ

[Tomas Pastore @tomasgpastore]

I kept leaving my laptop open overnight so an agent could finish.

So I built Murmello: a Mac terminal that opens Claude Code or Codex on its own cloud machine. Start it, close the lid, pick it up later right where it was.

Get early access → https://t.co/CwKdFA7Bmr