This is how a code factory should work. I was waiting at the doc’s office and filed 6 bugs - all from my phone. 1 is already merged, 2nd is merging, 3-6 are getting done in parallel. Using Symphony + Codex (Elixir, Codex App Server, Linear, GitHub)

Mine looks like this! - GH Issues + Linear w/ bi-directional sync - Auto-review from 3 different bots (gemini, claude, codex) requesting changes and triggering the locally running coding agents in real-time to provide those fixes. Integrates really well w/ Paperclip AI & OpenClaw, and even works for self-improvement loops too!

Nice

is this linear? i've been avoiding learning YET ANOTHER tool. but you keep posting all this cool stuffs damn you

Yup, just @linear :) Works mostly out of the box with Symphony

Do you have a tutorial on how you use symphony & linear? Would love to know more

i'll try to do a walkthrough soon

This is an interesting UI, but if using GitHub issues and PRs (which most developers are today) you can already do most of this in GitHub using the Agent control panel and using the agent of your choice running in parallel. FWIW, GitHub also has a mobile app :)

I built a similar setup with @droid (minus the video playwright directly in Linear) and let it run for 2,5 days. I checked in once a day in the morning to verify that bug bots hadn’t found anything critical. It had auto-merge enabled and could restructure tickets if needed to make them more “doable” during the build. 72 tickets shipped. @FactoryAI just one-shot it. Precision: 0 → 1 build. I spent a full day writing (and rewriting 😅) 12 spec docs and 1 droid mission in @chatprd before the build. It would have taken me at least 1 (or 2) weeks to achieve the same build with @cursor_ai. And we are still early. 😅

@mingjie is exploring few platform/API changes which would allow us to make this even better/faster

Ours looks like this ♥️

wwwow this is sick

:)

I love Symphony (except for token usage), but it's missing a critical step where I still want to be involved - discussing the implementation plan. Something I want to address soon (can have "Plan Feedback" status in Linear similar to Human Review)

do you think this can be ported to a claude code set up easily?

Did you do a ton of customization on top of Symphony to mitigate token consumption? I tried it out and it burned a few billion tokens overnight

I merged the optimizations that @odysseus0z proposed and it helped. I'm actually having codex review token usage as we speak to see if it helped.

@odysseus0z Nice, I’ll check that out!

Which mobile app is this?

@linear

Can you open source your implementation of symphony?

I'll probably do a video on it soon. Happy to share what I'm doing but need to get it a little more stable first

the best engineering teams dont optimize for lines of code shipped.\n\nthey optimize for feedback loop speed.\n\nfrom bug report to merge while waiting at the docs - thats not just efficiency. thats the compounding effect of good tooling.

Im checking your posts to stay updated with these types of workflows. Amazing work.

I need to see your setup! I set it up yesterday and it creates workspaces that just clone my whole repo not a big fan. Tinkering it to just do worktrees instead

How do you confirm the bug is fixed? Just by reading the PR or do you manually test every fix before merging?

so cool!

this is exactly what async dev workflows should look like filing bugs from a waiting room and watching them get merged in parallel — your team doesn't even know you weren't at your desk this is the actual productivity unlock nobody talks about. not 10x code, but 0-interruption async execution

Wait what's the mobile app?

Can't you just do all that from the Claude app? Connects with Claude Code/git, can also run it from the phone via /remove etc.

For acceptance criteria and validation do you type that out yourself? Automate that somewhat?

The fact that you can orchestrate parallel development from a waiting room shows how much tooling has evolved. Most companies are still stuck in the linear handoff model while you're running a distributed team from your phone.

that’s how you make the most of downtime. shipping fixes from anywhere feels great. love that workflow!

parallel fixes sound great but require near-perfect task breakdown to work. i’ve tripped over that when building, where the pre-work to parallelize took longer than just fixing one at a time.

Filing bugs from your phone and having them auto-merge while you're still in the waiting room. This is the future of software development — humans as architects, agents as builders. The parallelism is what makes it click

@grok how can i set up like this? what should I learn? any recommend resources?

I'm curious to follow this.

What do you wish was better about using your phone to build products? I also use Linear with agents funneling through to GitHub.

Di you opt for a second pro subs?

No, just bought $40 of credits

If u hit those 40 usd, the perk is switch to wifey or another chatgpt plus account it has more usage than the 40 usd, and u dont lose ur session 🙃

crazy how i can pass this video to openclaw and upgrade my own development pipeline with no additional instructions. Thanks for sharing. Super helpful

6 bugs from a doc's office and 1 already merged? That's the async dev loop in action. The real win isn't the tools - it's that you've built a CI/CD pipeline for your brain. Most folks spend weeks setting this up. You're shipping while waiting for antibiotics. 🚀

How about making your software remixable and everyone contributing to fixing/adding @RemixDotOne

interesting