Find out why players quit.
The User Interview Agent holds one-on-one interviews with hundreds of players at the same time, asking follow-up questions as each conversation unfolds. Every finding in the report is backed by players' own words.
- Players can start an interview any time
- 24/7
- Interview languages
- 15+
- From interviews to report
- <24h
Gear choice unclear after level 3
No nível 3 tentei várias vezes e não entendi que equipamento eu devia usar.
You ask the question, and the agent runs the interviews.
It drafts the guide, runs the interviews and prepares the report.
- 01
Ask your question
Describe what you want to find out in one sentence
- 02
Draft the guide
The agent writes and rehearses it
- 03
Choose participants
Select players by what they did
- 04
Run live interviews
Hundreds of one-on-ones at once
- 05
Get the report
Quotes and counterexamples behind every finding
The agent checks past findings first.
The agent answers from past findings and quotes in the knowledge base when it can. It adds every approved finding to the knowledge base.
Check existing evidence
The agent checks the knowledge base first. If the evidence is already there, you get the answer and skip a new round of interviews.
Interview players in Japan, Korea and Brazil in one study.
Players answer in their own language, by text or voice. Your team reads it all in one report, where each quote keeps the original wording and a translation.
Every player, their own language
Share of answers by language
- 31%
United StatesEnglish
I couldn't find anyone to team up with
- 18%
JapanJapanese
チュートリアルが長すぎました
- 14%
South KoreaKorean
보상이 너무 적었어요
- 11%
SpainSpanish
Me encantó, pero es un poco caro
- 9%
BrazilPortuguese
Adorei o visual do jogo
- 7%
GermanyGerman
Das Tutorial war zu lang
- 6%
FranceFrench
J'ai arrêté au bout d'une semaine
- 4%
TürkiyeTurkish
Oyunu çok sevdim
Players answer in their own language
Compare findings by market
Tutorial feels too long
Hear tone and emotion
“I mean… it's fine, I guess. I just stopped.”
Every market in one report
Answers in many languages, read together in one reportFollow-up questions find the reason behind a score.
You set the follow-up depth for each question, and the AI moves on when a player has nothing more to add.
How satisfied are you with this season's battle pass?
A 3.1 average gives no reason.
The XP wall after tier 12 is the top reason players stall on the battle pass
Every finding links back to what players said.
Each finding shows how many players said it, their quotes, and which players disagreed.
Just clearing my dailies takes forty minutes. After work I don't have time to play anything else.
Veteran players leave mostly because dailies take too long
Supporting quotes · 1 counterexample · 2 / 3
Approved by researcher- Follow-ups stop at the depth you set
- Every interview gets a quality score
- Findings come with quotes and counterexamples
- Only approved findings reach the report
- Triggered by behavior
- Players can opt out
- Frequency caps and quiet hours
- Keeps running
Let player behavior trigger invites.
Set a trigger for 7 days without a login or for leaving a bundle page without buying. Frequency caps and quiet hours limit invites, and players can opt out at any time.
- 10:02Cleared level 3
- 10:15Opened the first-purchase bundle
- 10:16Closed the bundle page without buying
“I couldn't tell what the stuff in the bundle actually does, and I didn't want to waste my money.”
Participant 147
Got 5 minutes to talk about that bundle? Your answers help us make the game better.
StartCheck what players say against what they did.
Select players by play history, and the AI uses that history in its questions. Findings are saved as user tags for building player segments.
“I still love the game. I've just been busy lately.”
Said vs. did · Returning players study
Logged in daily until the event ended, then dropped to once a week. Never opened the new mode.
- The new mode's entry requirement put returning players offConfirmed
- Event rewards felt weaker than last season'sUnproven
- The battle pass price is the main reason for leavingOverturned
Researchers can step in at any point.
Researchers can listen in, pause or take over any interview, and they approve every finding.
It drafts the guide, then rehearses it with simulated participants before you publish.
- A guide from a one-sentence goal
- Follow-up depth for each question
- Checks question order, length and wording that steers answers
- Rehearse, then lock the guide on publish
Describe your question, and the agent drafts the guide
- 1IntroWelcome and overview2 min
- 2Warm-upTheir last session3 min
- 3CoreWhen they wanted to quit6 min
- 4Wrap-upSummary check2 min
In your first week, when did you feel like stopping? What happened right then?
Get results in days, with interviews running in every time zone.
Players join from a link at any hour, and the agent writes the transcripts.
| Step | Traditional interviews | User Interview Agent |
|---|---|---|
| Recruiting and scheduling | 1 to 2 weeks of finding times | Players join from a link |
| Running interviews | 3 to 4 a day, about 2 weeks | Hundreds at once, any time zone |
| Transcripts and notes | About 1 week by hand | Automatic, organized by theme |
| Analysis and report | About 1 week | AI drafts it, then a researcher confirms |
Powered by heari.ai
heari.ai is ThinkingAI's AI user research platform. Use it on its own or connect it to Agentic Engine.
- Findings quote the players
- Interviews by text or voice
- From ThinkingAI, which works with 1,500+ enterprises
