Product Documentation

Petavue Skills

Skills are the expertise the agent applies to recurring revenue questions — the metric, the choices, the computation, and the limits that travel with every number. Same question, same method, every run.

What a skill is

A skill is the expertise the agent applies to answer one of those recurring questions the right way for your kind of business.

Ask the agent the question on its own and it answers from general marketing knowledge: plausible, but rarely precise to how your team defines a touch, a source, or a sourced deal.

Run a skill and the agent follows an explicit, pre-built method instead — the metrics, the choices it needs you to confirm, the computation, and the limits that travel with the number. Same question, same method, every run.

The agent does the work. The result is a dashboard you can trace back to the deal, with a methodology card that shows how every figure was built and Sage on hand to explain any number down to the source.

What's in the library

Five skills are live in this release. Each answers a question that used to take your team days to assemble, and each produces a dashboard.

What to expect when you run one

The first run of a skill is a setup pass. The agent prefills what it can detect from your data, you confirm or edit a few choices, and you approve the plan before anything builds. Nothing runs until you confirm. Building the dashboard takes roughly 15 to 20 minutes, depending on the skill.

After that, the skill replays your locked setup on a schedule and the number doesn't drift between them.

Set against the manual version — days of assembling and reconciling across systems, with no trail when the number gets challenged — that's the trade: a one-time build of around 15 to 20 minutes that you can defend line by line, then refreshes you barely have to think about.

What a skill won't do

Each skill is deliberately bounded. It answers one question well rather than every question loosely.

  • No invented numbers. If a required source isn't connected, the agent stops before running and names exactly what's missing.
  • No papering over partial data. If a needed field covers only some of your deals, the agent runs on what's there and surfaces the coverage gap above the number, so you know it ran partial.
  • No drift between runs. Same data and same answers produce the same result. If the number moved, the data moved or you changed an answer.

Next steps

  • New to skills? Start with How Skills work — the five-step run, start to finish.
  • Ready to run one? Pick from the library above. Marketing Influenced Pipeline runs on defaults if you want the shortest path to a first result.