Last updated: October 2, 2026
Short answer
An AI-moderated focus group is a qualitative study in which an AI system takes the moderator's job with real participants: asking the discussion guide, probing answers, and summarizing where people agree or split. In 2026 most tools sold under that name actually run many parallel one-on-one AI conversations and aggregate them. Few tools put real participants in one live room with an AI moderator.
Key takeaways
- "AI focus group" is used for at least four formats: parallel one-on-one AI interviews, live large-group text sessions, human-moderated groups with an AI co-pilot, and synthetic groups with no real people.
- Only formats in which participants hear one another produce group interaction data, which is the original reason to run a focus group.
- Published self-serve prices are rare. As of October 2, 2026, most enterprise vendors quote on request.
- The peer-reviewed evidence is thin and mixed: human moderation still produced the richest discussion in a 2025 experiment.
- Pick the format by your research question first, then the tool.
What is an AI-moderated focus group?
A focus group is a moderated discussion among a small group, usually 5 to 8 people, whose value lies in how participants react to one another (see what is a focus group). An AI-moderated focus group keeps the real participants but replaces or assists the human moderator with a large language model. The model asks questions, follows up, tracks the guide, and drafts a synthesis.
The label is now used loosely. One vendor, Perspective AI, defines an AI focus group in a May 4, 2026 guide as a study where an AI agent holds conversations with many real customers "in parallel, asynchronously" (Perspective AI, 2026). Koji's documentation describes the same design: many individual conversations that are aggregated afterward (Koji docs). That is a legitimate method. It is closer to a scaled interview study than to a classic focus group, because no participant ever hears another.
Four formats sold as "AI focus groups"
Before comparing tools, decide which of these you actually need. The format determines what kind of data you get.
| Format | Participants | Do participants interact? | Typical output | Best for |
|---|---|---|---|---|
| Parallel 1:1 AI interviews ("async AI focus group") | Real, each alone | No | Individual accounts aggregated into themes | Scale, reflection, sensitive opinions, multilingual fieldwork |
| Live large-group text session | Real, dozens to hundreds online at once | Indirectly, by voting on others' answers | Ranked responses, agreement levels, segment splits | Concept and message testing, prioritization |
| Human-moderated group with AI co-pilot | Real, 5 to 8 together | Yes, fully | Group discussion transcript plus AI prompts and summaries | Norms, shared language, academic focus-group designs |
| Synthetic focus group | LLM-generated personas | Simulated only | Simulated discussion and report | Piloting a guide, hypothesis generation, teaching |
For a deeper comparison of the first and third formats, see async AI focus groups vs live sessions. For the evidence on whether synthetic groups can stand in for people, see synthetic users vs real participants.
AI-moderated focus group tools in 2026: a dated roundup
The table below is based only on publicly available information on each vendor's own pages, checked on October 2, 2026. It is not a ranking, and features change quickly. Confirm details with the vendor before you buy.
| Tool | Format (as publicly described) | Public pricing | Source |
|---|---|---|---|
| Remesh | Live sessions "with hundreds of participants online at once"; participants vote on each other's responses; also async (Flex) and video formats | Not publicly listed (demo request) | remesh.ai |
| Koji | Parallel individual AI conversations, aggregated into group-level themes | Published: EUR 1 per qualified text interview, EUR 3 per qualified voice interview | koji.so docs |
| Outset.ai | AI-moderated one-on-one interviews in video, voice, or text; focus groups not mentioned on the home page | Not publicly listed | outset.ai |
| Recollective | "Conversation Task": an adaptive AI moderator running one-on-one dialogues with many participants simultaneously (launched November 2025) | Not publicly listed | DRNO, Nov 3, 2025 |
| Discuss.io | Live virtual focus groups led by human moderators, plus separate AI-moderated interviews and AI analysis agents | Not publicly listed | discuss.io |
| Qualitati | Human-moderated groups with a real-time AI co-pilot (Active Listener, focus-group mode); synthetic focus groups as a separate tool | Published: free tier with 30 monthly credits; paid plans from EUR 9/month | qualitati.com/pricing |
The pattern is clear. Live, small-group sessions in which an AI moderator talks to several real people at once are still uncommon in commercial tools. Most vendors chose parallel one-on-one conversations because they are easier to run, avoid scheduling, and remove dominant voices. They also remove the interaction that defines a focus group.
How much does an AI-moderated focus group cost?
A traditional focus group is expensive mostly because of people and logistics. Drive Research, a US market research firm, put focus groups at "$7,000 to $20,000+ per group" in November 2025, with participant incentives of $75 to $150 each (Drive Research, 2025).
AI formats remove the facility and most moderator time, but not recruitment or incentives. Where per-interview prices are published, the AI moderation itself is a small share of the budget; finding and paying the right participants dominates. Enterprise platforms generally quote per project, so compare total cost per completed participant, not list price. Our guide to AI user research costs breaks this down further.
What does the research say about AI moderators?
The academic evidence is still early. Three studies are worth knowing:
- Focus Agent (IVA 2024). Zhang and colleagues built an LLM framework that can moderate a focus group and simulate participants. They ran five sessions with 23 human participants and reported that simulated opinions resembled human ones (arXiv 2409.01907). It is a small feasibility study, not a validation.
- AI-assisted focus groups (Qualitative Research Journal, 2025). Chen, Wang and Chang compared human-moderated virtual groups with AI-assisted conditions. Human-moderated groups produced the most emotionally rich and contextually nuanced responses; AI summaries with human oversight kept comparable thematic depth, while AI-only summaries reduced engagement (Chen et al., 2025, DOI 10.1108/qrj-04-2025-0145).
- An AI bot that broadens discussion (2025). Vuk, Sarasua and Gilardi ran two preregistered experiments in which an LLM bot added missing arguments to online group chats. The range of arguments widened, even when the bot was disclosed as AI (arXiv 2506.17073). This supports one moderator job, countering groupthink, rather than full moderation.
Read together, the evidence supports AI as a co-pilot or analyst for group research today, and treats a fully AI-run live group as an open research question. We discuss the groupthink evidence in do AI focus groups reduce groupthink?
Qualitati's AI Focus Group Format Selector
This is a Qualitati-owned decision aid. Answer the questions in order and stop at the first "yes".
- Do you need to observe how participants influence each other (norms, shared language, disagreement)? Yes: run a live group with a human moderator and an AI co-pilot.
- Is the topic sensitive, or are participants vulnerable? Yes: use one-on-one interviews (human or AI-moderated with oversight), not a group.
- Do you need reactions from 50+ people to the same concepts quickly? Yes: a live large-group text session or parallel 1:1 AI interviews.
- Do you need fieldwork in several languages at once? Yes: parallel 1:1 AI interviews, with a native-speaking researcher reviewing transcripts.
- Are you still drafting the guide or the hypotheses? Yes: a synthetic focus group as a pilot, then real participants.
Then check the tool against this checklist:
- Does the vendor state plainly whether participants see each other?
- Can a human researcher observe and intervene during sessions?
- Are participants told the moderator is AI, and how is consent recorded?
- Where is data processed and stored, and is that compatible with your ethics approval and GDPR obligations?
- Can you export full transcripts for your own analysis, not only AI summaries?
- Is pricing published, or will you need a quote?
Where Qualitati fits
Qualitati is a European, budget-friendly qualitative research platform for universities and research companies, with GDPR and data privacy as central priorities. For group research it offers two tools, and we describe them precisely because the category is confusing:
- Active Listener (focus-group mode). A human moderator runs the live group. Qualitati transcribes in real time, suggests follow-up questions, tracks sections and time, attributes speakers after the session, and drafts a debrief. This keeps real group interaction while reducing moderator load.
- Synthetic Focus Groups. Multi-agent discussions among AI personas (up to 12 per session, optionally imported from consented Digital Twins) for piloting a guide or generating hypotheses before fieldwork. It is not evidence about real people.
For scale, Qualitati's AI-moderated interviews run one-on-one text or voice conversations in 10 languages, and ThemeLens analyzes up to 100 transcripts at once with participant-anchored quotes. Qualitati does not currently put several real participants in one live session with an AI moderator.
Limitations and methodological concerns
- Construct drift. If participants never interact, you cannot report group dynamics. Describe the method accurately in your methods section.
- Thin validation. Published evaluations of AI moderators in live groups involve few sessions. Treat vendor claims about depth as unverified until you pilot.
- Ethics and consent. Disclose the AI moderator at recruitment and session start, and check your ethics board's position before fielding with students, patients, or other vulnerable groups.
- Cultural nuance. Multilingual AI moderation still benefits from a native-speaking researcher reviewing transcripts.
Human-review note: methodological claims here summarize published studies as of October 2, 2026. Check them against your discipline's standards and your institution's research-ethics policy.
Who this is for, and when not to use it
For: research companies running concept and message tests, academic teams with multi-country samples, and PhD students who need many conversations on a small budget. Not for: studies whose research question is about group interaction itself, highly sensitive topics, or any project where the ethics approval specifies a human moderator.
FAQ
What is an AI-moderated focus group?
It is a qualitative study in which an AI system handles the moderator's role with real participants: asking questions, probing, and summarizing agreement and disagreement. Many tools run it as parallel one-on-one conversations rather than a live shared discussion.
Are AI focus groups the same as synthetic focus groups?
No. AI-moderated focus groups involve real people. Synthetic focus groups simulate participants with LLM personas and are suited to piloting and hypothesis generation, not evidence about real populations.
Which tools offer AI-moderated focus groups in 2026?
As of October 2, 2026, public pages show Remesh running live large-group text sessions, Koji and Recollective running parallel AI conversations, and Outset.ai running one-on-one AI interviews. Discuss.io and Qualitati support human-moderated groups with AI assistance.
How much does an AI-moderated focus group cost?
Most enterprise vendors do not publish prices. Where prices are published, AI moderation is cheap relative to recruitment and incentives. A traditional group typically costs $7,000 to $20,000+ according to Drive Research (2025).
Can an AI moderator replace a human moderator?
Not for every study. A 2025 experiment found human-moderated groups gave the richest discussion. AI works best today as a co-pilot, or for parallel one-on-one conversations at scale.
Bottom line
Choosing an AI-moderated focus group tool in 2026 starts with one question: do you need participants to interact? If yes, keep a human moderator and add an AI co-pilot. If no, parallel AI interviews give you scale, languages, and speed at a fraction of the cost of a traditional group. Use synthetic groups only to prepare for either.
Start free with 30 credits, no credit card required. Create an account to run an AI-moderated interview study or a synthetic focus group, view transparent pricing, or compare Qualitati with Outset.ai. Planning your guide? See how to write focus group questions and how many focus groups you need.