Displayr Skills: Encode Your Research Methods Into AI

Every research team has standards. The problem is they tend to live in the heads of the people who set them – which means the quality of any given piece of work depends on who happened to be on it.

Displayr’s new Skills capability is how you fix that.

What is a Skill?

A Skill is a set of plain-English instructions you save inside Displayr. The AI reads your Skills, recognises when one is relevant to what you’re asking in Chat, and follows its steps – or you can call a Skill directly by name. In effect, it lets you hand the AI your team’s playbook instead of hoping it guesses your house style correctly.

That matters because a general-purpose AI doesn’t know your standards: which variable counts as your market-share proxy, how you treat “don’t know” responses, or what a finished table should look like. Skills close that gap, so speed doesn’t come at the cost of the methodology your results depend on.

Skills are worth creating when:

1. You run the same task over and over.
2. Several people need to follow one agreed process.
3. Your organization has a house methodology it expects everyone to use.
4. You want the AI to produce consistent output rather than a slightly different result each time.

How to use Skills in Displayr

Finding the Skills menu

The Skills menu sits in Chat, just below the Ask anything… box. Open it to see the Skills you’ve built and can edit. Displayr also runs a library of pre-made Skills quietly in the background; those aren’t editable, but they’re already helping behind the scenes.

Creating a Skill

You can create a Skill in three ways:

1. Open the Skills menu, click New, and complete the fields.
2. Write a text file and save it to the Displayr Cloud Drive (covered under Going further below).
3. Ask Chat to build the Skill for you.

Whichever route you take, a Skill has four parts:

1. Name — lowercase, words joined by dashes, e.g. market-share rather than “Market Share”.
2. Description — a short summary of what the Skill does, read by both you and the AI.
3. When to use — the cue that helps the AI decide whether this Skill fits the request, including when someone describes the goal without using the obvious keyword.
4. Instructions — the steps to carry out. Precision here is what separates a Skill that works every time from one that occasionally wanders.

A worked example: market-share

Say you want a Skill that calculates market share from survey data. You’d name it market-share, describe it as calculating brand or company share using the best available behavioral proxy, and set “When to use” to fire whenever someone asks for market share, brand share, or share by brand — even if they don’t use the word “share”.

The instructions might tell the AI to:

1. Find the best market-share proxy in the study – last brand purchased, main brand used, or favorite brand – and ask the user if none exists.
2. Duplicate that variable set and name the copy “Market Share”.
3. Build a table from it, tidy any messy “Other” label, set non-brand options such as “don’t know” to missing, and sort the categories from highest to lowest with NET and Other pinned to the bottom.

That’s the whole pattern: name it, describe it, tell the AI when to use it, and spell out the steps.

Running a Skill

There are three ways to set a Skill running:

1. Pick it from the Skills menu — this drops a prompt into Chat, which you send by clicking the arrow.
2. Reference it in the prompt box and run it yourself.
3. Just describe what you want in Chat — the AI matches your request to the right Skill and runs it for you.

Where Skills are saved

Saving a Skill stores it as a .QSkill text file in a folder on the Displayr Cloud Drive. The Skill is then available whenever you’re editing documents in that folder or any folder beneath it — so the folder you choose decides who can reach it.

Going further with Skills

Once the basics click, a few more options are worth knowing:

1. Build Skills by hand. Create a .QSkill text file yourself and save it to the Cloud Drive. As long as the name, extension, and folder are valid, it shows up in the Skills menu.
2. Keep different versions per folder. When two Skills share a name, Displayr prefers the one in the same folder as your document, then works its way up to the home folder. A Skill saved at home is available everywhere, but a client folder can hold its own tailored version.
3. Have Skills call other Skills. Rather than copy the same steps into many Skills, point one at another — for instance, a brand-awareness Skill can finish by running an add-page-with-purple-bar-chart Skill to present its results.
4. Name the exact menu path. Displayr often has many ways to do the same thing, so spell out the route you want — e.g. Data Sources > + > Text Categorization — instead of leaving the AI to choose.
5. Run code when you need to. For technical users, a Skill can execute QScript via runHandCodedQScript(), for example to apply a visualization template to a new page.

Tips for writing Skills that work

1. Name them so the purpose is obvious. market-share tells you and the AI far more than analysis or john-1.
2. Begin small, then test. Start with the simplest version that could work and add detail only where the results call for it.
3. Only encode what you’re sure of. A wrong step blocks the AI from reasoning its way to the right answer — if you’re unsure, leave it out rather than guess.
4. Be specific. “Duplicate the selected variable set, or ask which one to use” is safer than “duplicate the data”.
5. Point at real menu options. Naming the exact feature path stops the AI picking the wrong tool from the dozens available.
6. Lean on Templates and QScript. Have a Skill apply a saved Template instead of rebuilding a chart, and convert heavily-used steps to QScript for consistency.

Frequently asked questions

Try it on your next project

Pick one task your team does the same way every time — a standard clean, a recurring table, a piece of analysis you always run — and turn it into a Skill. You’ll do that work once, and the AI will repeat it correctly for everyone from then on. Full step-by-step detail lives in the Skills and Advanced Skills help articles.