Define the study
Your research framing and your topic list are two sides of the same decision. The context and objectives you set here tell the AI Moderator what you’re studying and how deep to go. The topics you add tell it what to actually ask about. Getting both right — and making sure they’re consistent with each other — is the foundation of a well-designed conversation.
The quality of your output depends heavily on the quality of this input. Vague context produces shallow conversations; specific, well-considered framing is what gives the AI Moderator something to calibrate against when deciding how deep to go on a topic.
Research framing
Section titled “Research framing”These fields are in the Frame the Study section within the Conversation Setup panel (Panel 1). Think of them as the briefing you’d give a human moderator before fieldwork.
Context & Background
Section titled “Context & Background”Describe the background of your research project (maximum 1000 characters). Use this field to set the scene — who the participants are, what the research is about, and any relevant context the AI Moderator should know before the conversation begins.
If you go over the limit, the field shows Character limit exceeded.
Research Objectives
Section titled “Research Objectives”State what you want to learn from this research (maximum 1000 characters). If you go over the limit, the field shows Character limit exceeded.
Be specific. The AI Moderator uses your objectives to understand the purpose of each topic and the depth of inquiry expected. “Understand customer attitudes to the brand” gives it very little to work with. “Understand what drives brand trust among first-time customers, and whether that differs from repeat purchasers” gives it something to act on.
AI Moderator Tone
Section titled “AI Moderator Tone”Describe the personality you want the AI Moderator to bring to the conversation (maximum 200 characters). If you go over the limit, the field shows Maximum 200 characters allowed.
Examples: friendly and casual, fun and inquisitive, empathetic and respectful.
Topics
Section titled “Topics”Topics are managed in the Manage Topics panel (Panel 2). This is where you define the subjects the AI Moderator will cover during the conversation.
Each topic can be probed and prompted, which are two distinct phases:
- Probing — the AI Moderator follows up on topics the participant has already mentioned
- Prompting — the AI Moderator introduces topics the participant has not mentioned
See Design the conversation for how these phases work and how to write effective probes and prompts.
Adding topics
Section titled “Adding topics”There are two ways to add topics:
- Add New Topic — adds a blank row to the table for you to fill in manually.
- Extend Topics by Context — sends your current research context and existing topic list to the AI, which suggests additional topics and appends them to the table. A Generating topics… spinner shows while it’s working. All other topic actions are disabled during generation.
Topic table columns
Section titled “Topic table columns”| Column | Description |
|---|---|
| Topic | The topic name. Required, maximum 100 characters. |
| Description | Optional description of the topic. |
| Enable Probes | Toggle to activate probing for this topic. |
| Probes | Number of probe follow-up questions for this topic. Minimum 1, default 2. |
| Probes Override | Custom probe questions for this topic, entered one per line. If filled in, this completely replaces the Default Probe for this topic — the two are never combined. If left blank, the Default Probe is used. |
| Enable Prompts | Toggle to activate prompts for this topic. |
| Prompts Override | Custom prompt questions for this topic, entered one per line. If filled in, this completely replaces the Default Prompt for this topic. Only visible when prompts are enabled. If left blank, the Default Prompt is used. |
Each row has a delete action to remove that topic.
Reordering topics
Section titled “Reordering topics”Drag rows by their handle to reorder them. The order set here is the order the AI Moderator follows during the conversation (unless randomisation is enabled).
Randomisation controls
Section titled “Randomisation controls”Two toggles appear above the topic table:
- Randomise probe order — shuffles the order in which probe questions are asked across all topics.
- Randomise prompt order — shuffles the order in which prompt questions are used for topics that haven’t been mentioned.
Delete all topics
Section titled “Delete all topics”The Delete all Topics button removes every topic from the table, and is disabled when the table is empty.
Frequently asked questions
Section titled “Frequently asked questions”How much detail should I put in the Context & Background field?
As much as is genuinely useful to a moderator who knows nothing about your study. Include who the participants are, what the research is about, and any context that would help the AI Moderator understand the situation. If you wouldn’t bother telling a human fieldwork team, you can skip it here too.
Does the AI Moderator Tone actually change how the AI Moderator speaks?
Yes. Descriptors like empathetic and respectful or fun and inquisitive influence the register the AI Moderator brings to the conversation. Be specific — generic tones like professional are less effective than warm but curious or direct and energetic.
Do I have to fill in all the framing fields?
None are strictly required to save. That said, the AI Moderator works best with a well-defined research context and clear objectives — leaving these blank means it has less to go on when moderating.
How many topics can I add?
There’s no hard cap. In practice, keep the list focused on what’s genuinely essential to your research objectives — a long list of loosely relevant topics will spread the AI Moderator’s attention thin and make conversations feel like a checklist rather than a real exchange.
Does topic order matter?
Yes, unless randomisation is on. The AI Moderator follows the topic order you set here. Put your most important topics first in case the conversation runs short. Use the drag handle to reorder.
What happens if a participant mentions a topic that isn’t in my list?
The AI Moderator will only probe topics that are in your configured list. Anything mentioned outside of that list won’t trigger additional follow-up. If you want the AI Moderator to explore tangential territory, include a broadly-worded topic to cover it.
What’s the difference between Probes Override and the Default Probe?
The Default Probe (set in the Probing Strategy card) applies to every topic that doesn’t have its own override. A Probes Override on a specific topic completely replaces the Default Probe for that topic — the two are never mixed. See Design the conversation.
Can a topic be probed but not prompted?
Yes. Enable Prompts is a per-topic toggle — turn it off for topics you only want the AI Moderator to probe, not introduce. By default, prompts are disabled for new topics.
Why did “Extend Topics by Context” generate topics I don’t want?
The AI uses your current research context and topic list as input, so the quality of suggestions depends on how specific your Context & Background and Research Objectives are. After generation you can delete any suggestions you don’t want — use the delete action on each row or Delete all Topics to start fresh.
Can I update the research framing after a conversation is published?
Yes. You can edit any part of the builder after publishing. Save your changes to create an updated draft, then publish again when ready. Each publish creates a new version in the version history.
Common issues and solutions
Section titled “Common issues and solutions””My tone description doesn’t seem to be changing how the AI Moderator speaks”
Section titled “”My tone description doesn’t seem to be changing how the AI Moderator speaks””The AI Moderator’s tone is shaped by the combination of Context & Background, Research Objectives, and AI Moderator Tone working together. If the tone feels off, try being more specific in the description — for example, playful and conversational rather than just friendly — and test the difference in the conversation preview.
Also check whether Paraphrase is enabled on your probes and prompts. When paraphrase is off, the AI Moderator delivers probe and prompt text largely as written, which limits how much the tone setting can influence the conversation. Enabling paraphrase gives the AI Moderator room to vary its wording — and that’s where the tone description has the most effect. See Design the conversation.
”The Extend Topics by Context button is disabled”
Section titled “”The Extend Topics by Context button is disabled””This means topic generation is in progress. Wait for the Generating topics… spinner to finish — all topic actions (add, edit, delete, reorder) are disabled during generation.
”I can’t see the research framing fields or topic table”
Section titled “”I can’t see the research framing fields or topic table””Panel 1 (Conversation Setup) and Panel 2 (Manage Topics) are collapsible. If either panel appears as just a header bar, click the header to expand it.
”I accidentally deleted all my topics”
Section titled “”I accidentally deleted all my topics””The Delete all Topics action can’t be undone from the table — but if you haven’t saved since deleting, you can discard the change. Click CANCEL in the action bar and choose to close without saving. The last saved version of the conversation (including your topics) will be restored when you reopen it.