This is what work looks like for me now. An AI intelligent orchestrator laid out as a kanban board across projects. I run this on multiple computers, each project has specific workflow instructions (just like real projects have different requirements). This ships code like i do, with my 30 years of engineering practice in play. All tailored to each project. Also, written in #golang of course... github.com/digitaldrywood… It can use @OpenAI or @AnthropicAI models as backends for the LLM or @ollama. You can specify which effort level per task, and effort level per lane (merging usually doesn't need xhigh for example). Again, all this knowledge lives in the workflow for each project, so each project has it's own knobs to turn for efficiency and accuracy when it comes to gating effort and accuracy of the work tasks. I've been using this for months, and can't express how effective this truly is for real developers. This is NOT vibe coding, this is real engineering practice put into a real workflow/process.
