Skip to content

Platform · Prompts

The request that worked becomes the team's standard.

A prompt is the ready-made request: the text you already found works, saved with fields to fill in. Instead of sitting in the history of whoever found it, it sits in the library — private, with your team or with the whole company — and anyone can use it in chat without rewriting a thing.

Try it freeBook a demo
See how it works
Skyller · Prepare a prompt

Prepare a prompt

Prompt with variables

Prompt content

Create an outline to {{objetivo}}. Consider the needs of {{publico}}.

Demonstration · illustrative data

FIELDS TO FILL IN

the request has marked gaps; the person only completes what is missing

PRIVATE · TEAM · COMPANY

whoever creates it picks the reach, and private stays private

ONE CLICK IN CHAT

pick it, fill it in and the finished text enters the conversation

ARCHIVE WITHOUT LOSING

it leaves the everyday library and comes back from the archived tab

Where this shows up in your day

The right way to ask lives with one person only.

Every team has someone who figured out how to ask. That person becomes the bottleneck: everyone calls them, they rewrite the same text, and when they leave the knowledge goes with them.

Sales

The proposal that comes out different every time

Each person builds the proposal summary their own way. The format changes, an item goes missing, and the customer receives versions that do not look like they came from the same company.

The same format, always

Customer service

Answering a complaint without improvising

The tone, what can be promised and what has to be escalated change depending on who answers. Training everything again with every new hire is not an answer.

The agreement, written down

Whoever is starting

“I don't know how to ask the AI”

The person opens the chat, looks at the empty field and gives up. It is not a lack of willingness: it is not knowing where to start on a subject nobody taught.

Starts from a ready one

How it works

From a text that worked to a team resource.

The path has five steps and none of them requires technical vocabulary. Whoever knows how to do the work can turn their own way of asking into something the others use.

  1. 01

    Write it or ask for help

    Paste the text that already worked, or describe what you need and let Skyller draft the first version for you to review.

  2. 02

    Mark the gaps

    Whatever changes with each use becomes a field in double braces, with a name, a label, a filling hint, a default value and whether it is required.

  3. 03

    Organise it

    Title, description, categories and tags. They are how someone finds the prompt later, in the library's search and filters.

  4. 04

    Pick the reach

    Private, with invited people, or with the whole company. Whoever receives it comes in as a viewer or an editor; only the owner changes the reach.

  5. 05

    The team uses it

    In chat, the person opens the library, picks the prompt, fills in the missing fields and the finished text enters the composer.

Live proof

The person fills in three fields. The rest is already there.

A good prompt has two kinds of gap: the one only the person asking can fill, and the one the system already knows. The second disappears from the person's view.

  1. Starting point

    A customer-reply prompt, created by the customer service lead and shared with the team.

  2. What happens

    1. 01In chat, the person opens the prompt library and picks “Complaint reply”.
    2. 02A short form appears: the customer's name, what happened and the agreed deadline.
    3. 03Fields such as the company name and today's date come already resolved; the final text enters the composer ready to send.

  3. Result

    The request comes out the same every time, in the tone the company defined — and whoever is using it did not have to learn how to write prompts to get there.

What this scene does not promise

  • A prompt does not run on its own: it prepares the request, and the person is the one who sends it, in the conversation.
  • A prompt is neither an AI model nor a document template: it is the text of the request, not the engine behind the answer nor a ready-made file.
  • Sharing with the team does not open it to the whole company: the reach is chosen prompt by prompt.

Scene with a fictional company and customer; the field form and the automatic filling are the product's.

To evaluate calmly

What you can build, and who sees what.

Select a topic to see its capabilities, conditions, and limits.

Fields with a contract
Every gap has a name, a visible label, a filling hint, a default value and a required mark. With a required field missing, the request is not assembled — instead of going out half-written.
Some fields fill themselves
Data about the person, the company, the brand and the current date are resolved at use time. The person only sees and fills what genuinely depends on them.
Create it with help
Describe what you need and Skyller writes the first version, already suggesting the fields from the text. You review, adjust and publish — the final wording is always yours.

Real application

One way of replying, adopted by six people.

The gain is not the polished text: it is the difference between one person who knows how to ask and a whole team asking the same way.

  1. 01Customer service lead

    Takes the reply that worked last week and creates a prompt from it, marking what changes in each case.

    What changes becomes a field; what is fixed stays written in the text.

  2. 02Customer service lead

    Writes the label and the hint for each field, decides which ones are required and organises it by category and tag.

    It is the filling hint that lets a colleague use it without asking.

  3. 03Customer service lead

    Shares it with the customer service group as viewers and leaves one colleague as an editor.

    An editor adjusts the content; the reach stays with the owner or an administrator.

  4. 04Customer service team

    In chat, picks the prompt, fills in three fields and sends. Everyone replies with the same standard.

    The company, brand and date fields come already resolved.

  5. 05Customer service lead

    Comes back to the library, sees how many times it was used, adjusts the text and archives the request that fell out of use.

    The usage counter shows what the team actually adopted.

Comes ready

  • Fields with a label, a hint, a default value and a required mark, written in double braces
  • Automatic filling of data about the person, the company, the brand and the date
  • A library with search, categories, tags, a usage counter and an archived tab
  • Assisted creation, duplication and sharing with roles

The company decides

  • Which prompts are published to the whole company and which stay within a team
  • Who can edit each shared prompt, person by person or by group
  • How categories and tags are organised so the team finds what it needs
  • Which agents and AI levels a person may use together with the prompt

Questions from whoever decides

What people ask before adopting it.

Next step

Take the text you rewrite every week.

Paste it into the library, mark what changes with each use and share it with two people on your team. In a week the counter shows whether it became the standard. Or book a demo and build the first one with our team.

Try it freeBook a demo