@RhysSullivan https://t.co/HgsLnA8Hul, I have been building this from last 3 months. Support for native sdk and opencode as well. Donβt have stacked diffs with graphite at the moment but you can offload a lot of work.
My 5th SaaS is cooking.
We replaced 4 tools at our org, saving $20k/year and the last 30 days look like this: β 265 PRs merged β 1,105 commits to main β 351K additions / 53K deletions.
This is the most fun I've had building in years.
@NathanFlurry We have built PM + Support/Live Chat + Docs + CRM all under one tool internally to solve our own needs because of disconnection of these multiple tools. Everything is AI first.
If this is something you are interested, please send me a DM
@digitalshane_ I get all the things reviewed by codex 5.4 high/xhigh, sometimes involving qwen3.6 plus in the process for review get those refined in the PRDs then execute the plan.
But both models are nerfed they are not used to be like they were originally
Spent 30 minutes with Claude and Codex on a React problem. Both kept piling on useEffects.
Tried the same prompt with @Alibaba_Qwen 3.6 Plus via opencode β one-shot fixed it by ripping out the unnecessary useEffects and switching to useRef.
The correct, cleaner solution. First try.
No single model wins every time.