Can a prompt replace Productboard?

Build a prioritisation model you can defend, rather than a roadmap tool.

Candidates are rows. Each has a name, a short description, and a score per criterion — reach, confidence, effort, strategic fit, and whatever else you actually argue about. Criteria have weights that live in one configuration file. The composite score is computed, never typed.

The feature that makes this worth building: sensitivity. For the current ranking, show which criterion weight would have to change, and by how much, for the top item to be displaced. Prioritisation arguments are almost always disagreements about weights disguised as disagreements about items, and putting the number on it ends them faster than any chart.

Every score needs a note explaining it, and an item whose scores have no notes is marked unsubstantiated and excluded from the published ranking. Guesses and researched estimates otherwise look identical in a sorted list.

Link each candidate to evidence — request URLs, quotes, ticket references — and show the count of distinct sources next to the score, because one loud customer and twelve quiet ones produce the same row otherwise.

Publish a read-only page with the current ranking, the weights in force, and the date they last changed. Keep every previous version of the weights, so "we deprioritised this in March" is answerable.

Out of scope: issue-tracker sync, automatic capture from support tools, multi-team portfolios, and permissions. Export as CSV so the scoring can move into whatever the team really uses.