Skip to content
KB

What is ScatterSpoke

ScatterSpoke collects feedback from people and turns it into something you can act on. The idea it is built around is that a form is a bad instrument for finding out why — it asks everyone the same question and stops wherever the author stopped thinking.

So ScatterSpoke offers two ways to ask, and the difference between them is who decides what gets asked next.

A Static Survey is the familiar thing: you write the questions, everyone answers the same set. It is fast, comparable across runs, and unlimited on every plan including Free.

An Agent Survey inverts it. You don’t author questions at all. The AI interviews you first about the outcome you want, then interviews each participant turn by turn, following what they actually said. Two people can end up in completely different conversations because the agent probed different things. Agent Surveys are a paid-plan capability.

The thing that makes the rest work

Section titled “The thing that makes the rest work”

Whichever kind you run, every response keeps a transcript — the ordered record of what was asked and how it was answered. Structured answers, charts, and filters are derived from that transcript rather than replacing it.

That sounds like an implementation detail. It isn’t. It’s why an AI-generated summary in ScatterSpoke can be checked: the theme it named traces back to real sentences somebody typed. A system that only stored the aggregate would have nothing to show you when you asked “who said that, and what exactly did they say?”

What’s in a workspace

Section titled “What’s in a workspace”

Once you’re in, the sidebar splits into two groups.

Feedback holds Surveys, Templates to start from instead of a blank page, and — when retro boards are enabled for your workspace — Retros, the live retrospective boards.

AI holds Ask, where you put questions to your accumulated feedback in natural language and get answers grounded in it, and MCP, which connects the workspace to the MCP server so you can create surveys and read results from your editor.

Above the sidebar is the workspace switcher; below it, your own chip showing which plan the workspace is on. Quick find (⌘K) jumps to anything.

The two kinds of people involved

Section titled “The two kinds of people involved”

This is the distinction that surprises people most, so it’s worth stating early: the people answering your surveys don’t need accounts and don’t cost you anything.

A Seat Holder enters the workspace, builds surveys, and reads results. Seat Holders count against your paid seats.

A Participant is someone tracked so they can answer questions. They arrive through a link, answer, and leave. They never enter the workspace and never consume a seat.

You can therefore survey four hundred people on a plan with five seats. Core concepts works through why the model is shaped that way.

Where to go next

Section titled “Where to go next”