Hi HN, I'm Axew, and I've been building and using Launchpad Studio and thought I could OS since what started as a desire to optimize my throughput actually changed my patterns in a profound way and I haven't gone back to my previous patterns of working ever since.
I first decided to hack myself a harness back when Ralph was becoming the "hottest thing" in town and appreciated how it greatly increased my throughput. I didn't like how tedious it was nor the solutions out there to help. So Launchpad Studio actually started as a TUI wrapper to enable me create Ralph workflows faster. But I quickly outgrew that and kept wanting to optimize higher up the chain to increase my throughput to allow me do so much more concurrently. Launchpad Studio is an evolution of that and it has been nice to see some of the things discovered in that journey also go mainstream (workflows, inter-agent delegation, goals, inter-thread communication etc) all of which are first class citizens in Launchpad Studio. Some like "compaction-less" threads have enabled me use my agents like actual individuals day in day out.
To be clear, this is still early and certain things may be flaky or certain ideas/implementations still maturing. As mentioned this is highly influenced by how I have used it for different kinds of work/projects and I hope more people can influence it further
The Ask: The repo is linked above. I'd love for you to share some feedback. My hope is that some people on a similar journey with some interesting ideas could share these and contribute so that we do not have to re-converge to the same optimizations. Also because it is vendor and model agnostic we get to not have to switch every 3 weeks when a new capability shows up in one locked platform vs another.
Hi HN, I'm Axew, and I've been building and using Launchpad Studio and thought I could OS since what started as a desire to optimize my throughput actually changed my patterns in a profound way and I haven't gone back to my previous patterns of working ever since.
I first decided to hack myself a harness back when Ralph was becoming the "hottest thing" in town and appreciated how it greatly increased my throughput. I didn't like how tedious it was nor the solutions out there to help. So Launchpad Studio actually started as a TUI wrapper to enable me create Ralph workflows faster. But I quickly outgrew that and kept wanting to optimize higher up the chain to increase my throughput to allow me do so much more concurrently. Launchpad Studio is an evolution of that and it has been nice to see some of the things discovered in that journey also go mainstream (workflows, inter-agent delegation, goals, inter-thread communication etc) all of which are first class citizens in Launchpad Studio. Some like "compaction-less" threads have enabled me use my agents like actual individuals day in day out.
To be clear, this is still early and certain things may be flaky or certain ideas/implementations still maturing. As mentioned this is highly influenced by how I have used it for different kinds of work/projects and I hope more people can influence it further
The Ask: The repo is linked above. I'd love for you to share some feedback. My hope is that some people on a similar journey with some interesting ideas could share these and contribute so that we do not have to re-converge to the same optimizations. Also because it is vendor and model agnostic we get to not have to switch every 3 weeks when a new capability shows up in one locked platform vs another.