Follows the application you are in
Running nowReads which program you have shared and picks the course for it. One it has never seen gets a course written for it on the spot, with no setup from you.
Learn the software itself, with an agent that watches what you do and catches the step you got wrong as you make it.
The Skills agent is for learning software rather than a subject. Spreadsheets, design tools, editors, browsers, the internal system you were given on your first day: the thing standing between you and the work is usually the tool, not the idea.
It is application-agnostic, because it learns from the picture rather than from the program. It looks at the window you shared, works out what you are using, and holds what it sees against the step you are supposed to be on. Nothing is installed into the software you are learning, so the same loop works in a spreadsheet and in a design tool, and in the in-house system nobody outside your company has ever seen.
It takes work three ways, and the first of those runs today. Share the window you are working in and it coaches you through that program live, step by step, out loud. The two still being built are turning a deck or a document into a narrated course you can keep, and watching a whole class of shared screens at once.
Some of this runs today and the rest is being built. Each one says which.
Reads which program you have shared and picks the course for it. One it has never seen gets a course written for it on the spot, with no setup from you.
Watches for the result each step is asking for, and speaks at the moment you go wrong rather than after you finish. It stays quiet while you are working fine.
A total reached by typing, dragging or pasting all count as done, because the skill is the result and not one memorised route.
Being built: demonstrating the action on screen for when describing it is not landing.
Being built: upload a deck or a document and get a narrated lesson back, section by section.
Being built: following every shared screen in a session and stepping in where a learner has drifted.
Learn a program by using it on real work, corrected as you go rather than after you have finished.
Turn the deck you already have into a narrated course, and watch a room of shared screens at once.
Give a new starter a patient second pair of eyes on the internal system nobody has time to sit and explain.
The same four steps on every Intelligenfy agent.
The learner selects a market, topic, strategy or skill they want to understand.
The learner deliberately shares a selected screen, browser tab or application with Intelligenfy.
The AI agent observes the screen you shared, explains what is visible and answers by voice or text.
The learner completes simulations, quizzes and guided exercises while Intelligenfy tracks progress.
More detail on how screen-aware agents work.
The others teach a subject while you work. This one teaches the tool you are working in, which is a different job: the lesson is the sequence of actions, not the ideas behind them.
No, and that is the point of the design. It reads the shared window itself rather than hooking into the program, so something it has never seen still works: it writes a course for it the first time somebody opens it.
No. Nothing is watched until you share a window and press Coach me, and you can stop at any moment. Pictures are read as they arrive and never stored. That rule is the same for every agent here.
Share a window in the workspace and press Coach me. It works out which program you are in, works through the steps with you and corrects you as you go. Courses from uploaded documents and live class monitoring are still being built.
It is not open yet. Join the list and you are invited before general release.