Hey Product Hunt, Furkan here. I built GoalT because every planning tool I used made me lie about my own project. They all assume a tree: one goal breaks into sub-goals, each sub-goal has exactly one parent. Real work isn't shaped that way. A single bug fix turns out to be load-bearing for three separate initiatives, for three unrelated reasons, and a tree makes you either duplicate that node or pick one parent and pretend the others don't exist. So GoalT is a DAG instead. Each parent splits exactly 1.0 of value across its children, and a goal with several parents accumulates from each one. The practical effect is that whatever more of the project quietly leans on rises to the top by itself, without anyone deciding it should. The number is a pull-weight, how much is riding on this, not a slice of a fixed pie. Value is conserved per parent rather than globally, so children can sum to more than the root. That confuses people at first and it is deliberate. It also runs live inside Claude Code: the dashboard highlights whichever goal is being worked on right now, marks goals whose files have uncommitted changes, and opens the diff in VS Code when you click one. What it is not: proven. There is no benchmark against AHP, ANP or ordinary weighted scoring yet, and no convergence result for LLM-weighted graphs beyond small ones. Both are written up as open questions rather than things I am hoping nobody asks. Runs in your browser with nothing installed: https://colab.research.google.co... Or: pip install goaltree. Apache 2.0. The question I would genuinely like answered today: point it at a project that is currently confusing you, and tell me whether the shape it produces is one you recognise. If it isn't, I want to know why not.

GoalT

See which goal is actually load-bearing, not just urgent!

See which goal is actually load-bearing, not just urgent!