Pretty neat concept, but doesn't seem to pick up more complex pipelines.

I tried "build and test both arm64 and amd64, deploy amd64 to staging and then run e2e tests" and just got a simple 4-stage labels.

I didn't expect it to work, and this was admittedly a bit of a tough example. But if you advertise Natural Language...

I built this using openAI codex, google cloudrun and streamlit (visualisations by graphviz - codex is not bad at translating pipelines to graphviz without really much prompting which somehow surprised me). Let me know what you think. No fine tuning required for this (as Codex already knows quite a lot).