An interactive is a live, working mini-app that runs inside your session. A poll that fills in as people vote. A quiz that knows who answered. A sorting activity that remembers where everyone dropped their card. Or a whole slide deck where the slides are the activities.
You don't build it. You describe what you want, and it exists — usually in under a minute, in the room, while people are still talking about it.
Interactives are in beta. Authoring is limited to an allowlist right now, but anything already built works for every participant.
The one-line version
Cards frame the activity. Interactives are the activity.
A board made entirely of HTML cards is a gorgeous, dead slide deck — which is exactly the thing Kolab is trying to get away from.
Creating one
Interactives are created and edited through Kola or the MCP connection — not from the toolbar. Describe the activity in plain language:
- "A spectrum where everyone drops a dot"
- "A five-question quiz on this"
- "A poll on slide 4"
Describe the activity, not the layout. "A sorting game where people group these ideas into three buckets" works better than "a three-column grid with draggable boxes."
Revisions work the same way — "make the options bigger", "add a slide", "change the question."
If the tool doesn't seem to exist: disconnect and reconnect your MCP server. It may not know about the tool yet.
Two ways to use one
| Placement | What it's like | Use it when |
|---|---|---|
| Full-screen | You activate it, everyone sees it, you drive | You're presenting — this is what replaces the deck |
| On the board | It sits on the canvas as a station people gather around | Four tables, four live activities, proximity audio between them |
Board placement is the better default for anything people should be able to return to. Full-screen is for the moment you want the whole room looking at the same thing.
Turn a deck into something people do
This is where full-screen interactives earn their place.
Think about what you actually screen-share: almost always a slide deck. Screen sharing makes that deck visible to everyone, but it doesn't change what it is — people watch, you talk, and knowledge moves one way.
An interactive deck is the same material with the passivity taken out. Slide 4 isn't a picture of a poll — slide 4 is the poll. People answer it, the results fill in live, and then you move to slide 5.
What changes
| Screen-shared deck | Interactive deck |
|---|---|
| People watch | People answer, vote, sort, choose |
| You ask "any thoughts?" into silence | The room's input is already on screen |
| Discussion happens (or doesn't) in chat | Responses are captured and stay with the session |
| The deck leaves with you | What the group produced stays in the space |
How to convert one
Kolab doesn't have file upload yet, so you can't hand it a .pptx directly. The deck's content needs to reach your AI client instead — paste in the outline, or attach the file there if your client can read it. A plain text outline is genuinely enough; interactive decks have been built from nothing more than that.
Then ask for what you want:
"Here's the outline for my onboarding session. Build it as an interactive deck — make slide 3 a poll on where people are starting from, and slide 6 a sorting activity for these six priorities."
Then iterate the same way you would with anything else: "add a slide", "make slide 5 a word cloud instead", "change that question."
You don't have to convert the whole thing. The highest-value move is usually picking the two or three moments that should have been participatory and making just those interactive. A deck where three slides are live activities beats a deck where all twenty are mildly clickable.
Beyond polls: slides can be word clouds, quizzes, sorting and matching activities, agree/disagree spectrums, even adventure-game branching. With some exceptions, if it can be done on the internet, it can be a slide.
How participation works
- Everyone's input syncs. Contributions appear live for the whole room.
- Late joiners catch up automatically. No need to restart or re-explain.
- Nothing is lost when you move on. Votes and answers stay attached to the session.
- You get a facilitator view. Interactives can show a facilitator bar that participants don't see — controls that are yours alone.
- You can reset. A facilitator-only reset clears the current round without destroying what came before.
Interactives vs. HTML cards
These ship together and are easy to confuse.
| Interactives | HTML cards | |
|---|---|---|
| What it is | A live, working mini-app | Beautiful static content |
| People can... | Vote, answer, drag, explore | Read it |
| Good for | Polls, quizzes, sorting, games, live decks | Instructions, a legend, a welcome block, activity setup, presenter bios |
| State | Syncs across everyone | None — it's the same for everyone |
Cards make a board look designed rather than assembled — real typography and layout instead of text boxes on rectangles. They're genuinely useful. They're just not the activity.
Note: "HTML cards" is a working name and is likely to change — it collides with Card Decks, which are a separate feature.
What people have built
| Interactive | What it does |
|---|---|
| Build a Duck | A LEGO® SERIOUS PLAY® inspired warm-up — everyone builds, then shares |
| Five Sensing | A grounding activity that collects what everyone noticed, with hearts that darken as they accumulate |
| Meet Kolab: Take a Look Around | An interface tour with clickable features that light up and warm as people explore them |
| Escape room | A multi-stage scenario built across a series of interactives |
| A converted workshop deck | Built straight from a plain text outline, with the passive slides turned into live activities |
Beyond those: polls, word clouds, quizzes, sorting and matching activities, and agree/disagree spectrums — either standalone or as slides inside a deck.
Rule of thumb: with some exceptions, if it can be done on the internet, it can be built as an interactive.
Working well in beta
Test before the session, not during it. Build it, click through it as a participant, then run it live.
Batch your edits. Every change republishes the whole interactive, so ten small tweaks cost ten full rebuilds. Collect your changes and make them in one pass.
Keep it lean. Very large interactives get less reliable well before any documented size cap. Simpler builds publish faster and more dependably.
Ask for vector, not screenshots. Interactives drawn as vector shapes stay crisp at any board zoom, and individual regions can change colour as people interact with them. Embedded screenshots can't do either.
Known beta limits
These are real today and being worked on:
| Limit | What happens |
|---|---|
| No duplicating | You can't yet copy an interactive into another space |
| Templates don't carry them | Saving an interactive as a board template and opening it in a new space doesn't work |
| No template preview | A template containing only interactives saves without a preview thumbnail |
| No preview before publishing | Every version goes live on the board — there's no draft or dry-run state |
| No file upload | You can't hand Kolab a .pptx or PDF — the content has to reach your AI client as text or an attachment there |
| Authoring is allowlisted | Only approved accounts can create them; everyone can use them |
Found something else? The most useful thing you can send is the activity you wanted to run and couldn't.
Where the name comes from
Interactive is a noun here — the way the world ended up with Shorts and Reels. An interactive. Several interactives.
Newsrooms have used the word this way for over a decade, and they mean what we mean: a made, self-contained, playable thing.
Related Guides
- Connect via MCP - How to reach interactives from Claude
- Tools vs. Canvas Elements - Where interactives sit among everything else
- Card Decks - A separate feature, despite the name collision
- Screen Sharing - What full-screen interactives replace
- AI Agenda Builder - Plan a session with Kola