Feature Prioritisation
The backlog has ninety items in it. Eleven of them came from a single loud customer, six came from a board meeting, and nobody remembers why the rest are there. When planning comes around, the ranking is a negotiation: whoever has the freshest anecdote wins. The evidence that would settle it exists, but it is scattered across Zendesk, Gong recordings, product analytics, and a spreadsheet someone maintained until March.
Skynet holds all of that in unified memory and applies your criteria to it. Ask why a feature is ranked third and it shows you the seventeen tickets, the two deals it blocked, and the goal it maps to. The argument stops being about who feels strongest and starts being about what the evidence says.
How it works
Connect the evidence
Point Skynet at your support tool, your CRM, your analytics, and your feedback board. It builds one memory across all of them, so a feature request raised in a sales call and the same request filed as a ticket resolve to the same thing rather than two entries.
Define what counts
Tell the agent what your criteria actually are — reach, revenue exposure, strategic fit, effort. These are your weights, not a generic framework. Change them for a growth quarter and the ranking changes with them.
Score with the reasoning shown
The agent ranks each candidate and writes down why: the demand it found, where it found it, and which goal it serves. Nothing scores highly without a citation you can open.
Review and decide
You get a proposal, not a verdict. Override anything — the agent records the override and applies it next cycle. The judgment stays yours; the evidence-gathering doesn’t.
Build it from a prompt
You describe how your team prioritises once, then run it before every planning cycle.
What you get is a ranked backlog you can defend in a room. When someone asks why their feature is at number twenty, the answer is a list of sources rather than a shrug — and if they have evidence you missed, you add it and rerun.