Have in-depth conversations with 100 people in 1 week
The scale and speed of forms, with the depth of human conversations.
Your 24/7 research team
Run qualitative research at scale, using AI end-to-end in the process. The final decision is still yours, at every step.
Script assistantHi! I'll build the script with you. How would you like to start?
How do you want to start?
‹ ›Script creation
Describe your research goal and an AI assistant builds the script with you — from scratch, from a context file, or from a template.
Interview moderation
Research-specialized AI agents conduct voice interviews with your users and collaborators.
Analysis assistantWhat themes came up most in the interviews?
Using Transcripts · Audio
Results analysis
Insights extracted and synthesized in real time. You audit and make decisions 10x faster. Chat in natural language with the results.
Solutions
Product, Marketing and HR use Grunn to research in depth and at scale — without a scheduling bottleneck.
Continuous discovery, validation and testing with real users.
Churn analysis
Concept testing
Problem discovery
Product and feature launches
Usability testing
Whoever uses it decides better
The experience was very positive. We found a new way to get data faster and deeper — with real depth on the themes we cared about.
We managed to cover a huge number of people very quickly, in four days. If we had interviewed them ourselves, it would never have been possible. With a form, adoption would probably have been lower. It was absolutely worth it.
A qualitative study here normally runs with 6 to 8 people. With Grunn we scaled that number exponentially — we reached around 40 responses. I had never done that in my life.
Instead of a rigid script, the conversation flows naturally and warmly, adapting the questions to each employee — it prompts deeper reflection and more authentic answers than traditional surveys.
It was an 8-to-10-minute survey with real depth. I feel totally comfortable using it again — I'm actually excited for the next one.
We got a speed of responses I don't think we could ever have reached with a form. It turned out great. I would use it again.
We had never run a qualitative study interviewing 37 people at once. It brings a lot of insight.
Team adoption was excellent, and the voice format was the highlight: without the barrier of formal writing, the conversation flows far more naturally and engaged.
Most people really liked it — they found it different, really nice to talk to. It's more fluid than writing, which can feel a bit stiff. The experience was very positive, and I would use it again.
I saw a company with dedicated founders, eager to learn and very agile in every exchange. Long live Grunn, and thank you for the partnership!
The AI conducts the interaction with empathy, makes connections with what the person just shared and creates a sense of genuine listening. That care in the way it was conducted made a difference in the employees' experience — they reported feeling at ease to talk and to go deeper into their perceptions.
Scale or depth? Both.
Traditional interviews achieve depth; forms achieve scale. Grunn was built to deliver both at the same time.
| Grunn | Traditional interview | Form | |
|---|---|---|---|
| Depth | Dynamic follow-ups based on what the person said — gets to the why | High, moderator-dependent | Shallow — fixed answers to fixed questions |
| Scale | Hundreds of interviews in days | About 10 people a month across the full process | High, but without depth |
| Time to results | Real time — results right after each conversation | Weeks, across recruiting, conducting and analysis | Fast, but shallow |
| Evidence | Every conclusion linked to the source verbatim and audio | Depends on manual recording and note-taking | A number, without the context behind it |
| Cost | Per completed interview — a fraction of the traditional cost | High: recruiting, moderator and analyst | Low |
Enterprise-grade compliance
Privacy
LGPD compliant
AI models don't train on your data
Deletion on request
Infrastructure
Microsoft Azure
Data hosted in Brazil
99.9% uptime SLA
Inherits ISO 27001 and SOC 2 from Azure
Identity and access
Single Sign-On (SSO)
Role-based access control (RBAC)
Row-level security
PKCE + encryption, CSRF protection
Traceability
Activity logs
History of every action
Interview export
Periodic external security audits
Featured cases
CASE95 people interviewed in 5 days
Qualitative research at scale and depth — at the traditional pace, companies interview about 10 people a month.
View case →
CASE115 employees heard in 4 days
To redesign its technical career track, Sensedia heard 115 people in depth in a single week — at the traditional pace, interviewing them would take months.
View case →
CASEQualitative research at scale, for the first time
iFood Pago's design team traded weeks of recruiting for hours of real conversation — without giving up depth.
View case →
CASE65 employees heard about purpose
Culture and purpose are hard to capture in a form. Fretadão set out to hear, in depth, how its team sees the company's purpose.
View case →
CASEEmployee climate research by voice, in depth
In-depth voice-AI climate research, with 89% employee approval.
View case →Frequently asked questions
Does the AI reach the same depth as a human researcher?
Yes — because it doesn't read a fixed script. Grunn's AI asks follow-ups based on what the person just said: it asks for examples, probes contradictions, pulls the why. The script works as a guide of objectives, and the moderation was designed with qualitative research technique — it is not a generic chatbot.
How does Grunn guarantee the AI doesn't make up conclusions?
Every insight in the report is traceable to the literal excerpt of the interview that originated it — you can click and listen to or read the source. Anything without grounding in what the respondent said does not enter the output. The methodology was built with a trained researcher, separating what was said from what is interpreted.
Is the research anonymous? What about data protection?
Yes. There is explicit consent on entry and a recording opt-in, and results can be delivered aggregated and de-identified. The LGPD framework covers consent, security and data hosted in Brazil.
Doesn't the AI voice sound robotic?
The interview is by voice and in real time, with natural turn-taking. The AI adapts its pace to people who say little, welcomes and redirects those who go off-topic, rephrases questions in context when misunderstood and moves on when a topic has been covered.
How much does it cost?
The model is per completed interview: you pay for the finished conversation, not per seat or license. Compared to traditional qualitative research — recruiting, moderator and analyst — the cost per interview is a fraction. For high volume, there are tiered packages.
What is the response rate?
Being asynchronous and voice-based, the barrier to participate is low: people answer whenever they want, by speaking. The rate depends mostly on how the invitation is distributed — channel, approach and timing — and, in validated cases, it came out above traditional surveys.
What is the ideal interview length?
Focused interviews, usually under 15 minutes, with few objectives per round. Because the AI probes dynamically, a few well-moderated questions yield more than a long questionnaire. For time-poor audiences, short recurring pulses work better than a single long interview.
How long does it take to get started?
A matter of days: building a script and going to field takes up to about a week, and the learning curve is fast. Distribution is via a shareable link — carrying your company brand — and, for continuous enterprise volume, there is an API integration path.
Aggregate metrics from research run on the platform · data as of July 2026
Start interviewing today
Run hundreds of voice interviews in days. The first 25 are free.