ScatterSpoke is an AI survey that writes the questions and interviews every person, asking follow-up questions on what they actually said, then gives you the answer. Your first ten agent surveys are free.
Most AI survey tools stop once the form is written: people answer exactly the questions on the page and nothing more. Spoke writes the questions too, then listens to each answer and decides where to dig.
A form would record “fine” and move on. Spoke asks the one question that answer calls for. Read more in the probing questions guide.
The offsite made decisions nobody recorded, so they get reopened. The fix is notes, not another offsite.
Most surveys collect numbers or vague sentences. ScatterSpoke runs a real conversation with every person, then does the synthesis you'd otherwise spend a week on.
Type it the way you'd say it to a colleague. Spoke, the ScatterSpoke interviewer, turns it into a goal, an audience and the topics worth covering.
What do you want to learn, and who's answering?
Drafted the spec. Should I ask what they switched to, or keep it to their experience with us?
Spoke asks, listens, and follows up on what each person actually said. Each person gets a link by email or QR code and chats for the time you set. No account needed. Anonymous on paid plans.
Price barely comes up once Spoke follows up. Most churned customers describe the same wall: a data-source setup step that needs a credential they don't own, with the trial clock running. Several say they never saw the product working with their own data.
It reads what someone just said and decides what to ask next: pressing on vague answers, chasing the surprises, skipping what doesn't apply. You set how deep it probes each topic. No branching logic to build.
Open answers cluster into themes as they land, each with the quotes that prove it and the number of people who raised it. Your week of hand-tagging is gone.
When two groups explain the same problem differently, the answer shows both sides and how many people are on each, instead of averaging them into one number.
Classic surveys stay unlimited on every plan. Agent surveys use your plan's monthly allowance, and Free comes with ten to try.
You give it a goal, an audience and the topics you care about. It drafts the interview, then talks to each person on their own, following up on what that person said instead of walking everyone through the same fixed questions. Two people can get entirely different follow-ups.
Yes. Before anything goes out you see the survey spec: the goal, the audience, the tone, the topics with how hard to probe each one, and the topics that are off limits. Edit any of it. Respondents never see the spec; it only guides their interview.
Yes. Say what you want to learn and who is answering, and Spoke drafts the survey spec: the goal, the topics and how deep to probe each one. You edit anything before it goes out.
An agent survey can. Spoke reads each answer and asks the follow-up it calls for, so a vague “it was fine” gets a specific question back. Most AI survey generators only write a fixed form.
No seat and no access to your workspace. They get a link and the conversation happens in the browser; some settings ask them to verify an email address. Invited people who use Claude can also answer through their own assistant over MCP once signed in.
Every transcript, plus the Agent Results Brief that builds as responses arrive: a summary, themes with the quotes and the number of people behind each one, and the places where two groups disagree. Export the survey data as CSV whenever you want.
Your choice per survey. Anonymous responses are part of the paid plans; see pricing for what each plan includes.
Free to start. No card. Your first ten agent surveys are on us.
