Anyone can ask the AI for a report once. The reason to package it as a Skill is repeatability — the same shape, week after week, with whatever the latest data shows.
3 things to walk away with
Same shape
A Skill locks the shape of the output so every run looks the same — only the data changes.
Defined once
You codify what matters once; the Skill runs over and over without re-explaining.
Living data
The Skill always reads fresh source data, so the report reflects what is true today.
The Hiring Pipeline Use Case
The demo opens with a SharePoint site holding a candidate list and an open-roles list. The user types one sentence — "generate a hiring pipeline report" — and the AI returns an interactive HTML dashboard with open roles, active candidates, average days in stage, a Sankey flow, and a per-role breakdown.
This is not a wizard. It is a Skill the user defined earlier, and they can invoke it at any time by name. Quarter after quarter, the same report comes out — with the latest pipeline numbers folded in.
A real KB use case looks like this: messy source lists in SharePoint, a single sentence to the agent, and a polished interactive dashboard back. The agent does the assembly.
You need a BI tool or a Power BI license for this. The output is a self-contained HTML file generated by the AI — no separate dashboard product required.
What an AI Skill Actually Is
A Skill is a named, reusable capability you teach the agent. It looks ordinary — you describe what you want, what data sources to read, and what shape the output should take. Once defined, the Skill becomes part of the site's agent capability set: anyone with access can invoke it.
The demo's hiring report is one Skill. The author could also add a sales-pipeline Skill, a vendor-renewal Skill, a quarterly OKR Skill — each is a unit of reusable agent work.
Skills are building blocks. Each Skill turns a complex, multi-step agent workflow into a one-line user invocation.
A Skill is code you write in a programming language. It is a definition the agent reads — written in plain language with structure for the data sources and output shape.
The Skill Definition — what matters, what to read, what to produce
Inside the hiring-pipeline Skill, the user has called out specific pieces: the header, the metrics bar, the Sankey diagram, the exact pipeline stages to include, and the exit nodes. These are the things they want their leaders to see — and codifying them once means the Skill produces the same view every time.
The Skill does not constrain the agent to dumb templates. It still lets the agent reason about the data, choose appropriate chart types within the spec, and surface "at risk" candidates. But the overall shape — what counts as the report — is locked.
Defining the Skill is defining what matters to your leaders. The agent fills in the data; you define the structure that makes the data useful.
A good Skill leaves everything up to the agent. The opposite — a useful Skill is opinionated about structure so consumers know what to expect each run.
Repeatability — the actual reason to make a Skill
The author explains the primary motive directly: "I want to be able to ask for the same hiring report quarter over quarter and have it actually deliver something highly valuable to me." That is repeatability. Without the Skill, every quarter would require re-explaining the format, the metrics, the chart types, the at-risk thresholds.
With the Skill, the structure is committed. The next quarter, the data may be totally different — but the report will look the same so leaders can compare directly.
Repeatability is comparison-readiness. Locked structure across runs is what lets a viewer see "this quarter vs last quarter" at a glance.
Repeatability means freezing the data. It means freezing the shape — the data always changes; only the layout, metrics, and presentation stay the same.
Flashcards — Foundations
01 · Foundations
Hiring Pipeline Use Case
tap to reveal →
The demo scenario: messy SharePoint lists of candidates and open roles, a one-sentence invocation, and a polished interactive HTML dashboard back — assembled by an AI Skill the user defined earlier.
← tap to flip back
01 · Foundations
AI Skill
tap to reveal →
A named, reusable agent capability defined in plain language. Turns a multi-step workflow (read data, compute, render dashboard) into a one-line invocation any authorized user can call.
← tap to flip back
01 · Foundations
Skill Definition
tap to reveal →
The author's spec inside a Skill: data sources, metrics that matter, output shape (header, metric bar, charts), and explicit stages or thresholds. Opinionated about structure, flexible about data.
← tap to flip back
01 · Foundations
Repeatability
tap to reveal →
The actual reason to package work as a Skill. Locks the output shape so quarter-over-quarter comparisons are visually direct. The data changes every run; the structure does not.