In active beta · first project free

Run 50 user interviews this week. Without scheduling any.

You write the goal. An AI agent runs the conversations in WhatsApp, Telegram, or web chat — voice or text. You get transcripts and a decision-ready report. The depth of an interview at the scale of a survey.

In active beta. First project is free in exchange for product feedback.

Acme Research
via WhatsApp · online
10:24
Hi! We're trying to understand what made the first weeks of trial hard. Got 5 minutes?
yeah sure
Cool. When you signed up, what were you trying to get done that week?
0:24
You mentioned the close was painful. Which part specifically?
Message

How it works · no scripts, no branches to wire up

Say you want to know why customers cancel in their first 60 days.

01

You set the goal

~5 min

One sentence — the decision you're trying to make. The agent designs the conversation around it: opening, key questions, follow-up logic, scoring rubric. No scripts to write, no branches to wire up. You review and approve before it ships. Optionally, upload extra context (knowledge base, docs, past research) so the agent is fully fluent in your domain.

02

Generate the link

One link, branded with your logo. Respondents open the chat in their browser, or in any connected messenger — WhatsApp, Telegram, Slack, and more. Zero install. Zero login.

03

They chat with the AI agent

Mostly on their phone. Voice or text. The agent transcribes either, asks follow-ups based on what they said, skips what's already covered. Average completion: 8 minutes.

04

You get the report

Top findings ranked by impact, with verbatim quotes. Full raw transcripts exportable as JSON or CSV. Seven days from launch to decision.

No calls scheduled. No interviewers hired. No dashboards to build.

What your users see

Not a survey. Not a chatbot. A short, focused conversation.

Acme Research
WhatsApp · AI interviewer
today · 10:24
Hi! We're trying to understand what made the first weeks of trial hard. Got 5 min?
yeah sure
Cool. When you signed up, what were you trying to get done that week?
0:24
You mentioned the close was painful. Which part specifically — consolidations, intercompany, or the last-minute adjustments?
mostly intercompany. our entities aren't clean
Got it. Has anything you tried so far helped, even partially?
01

They already use it.

The conversation happens in WhatsApp, Telegram, or web chat. No install. No login. No new app.

02

Voice or text — their choice.

Most respondents switch between both. The agent transcribes voice in any of 30+ languages. Even on a noisy train.

03

The agent only goes deeper where it matters.

No 12-page form. No mandatory fields. If they answered something already, the next question knows. If they said something interesting, the next question asks why.

Beta data: 30–40% completion rate on warm lists. Average finish: 8 minutes.

What you get back

A report, an analyst, and the raw data.

One option is the auto-generated Snapshot. Another is the chat with your data. The third is to take the transcripts and run your own analysis. Most teams use all three.

Snapshot — Churn discovery
Trial → paid · 187 respondents
Auto-generated
1
Setup friction in the first session
"I couldn't get our data in without a call with support"
73 of 187
2
Unclear value before the second login
"I wasn't sure what I was supposed to do next"
44 of 187
3
Missing integration with our stack
"we already use three other tools — this didn't plug in"
29 of 187
Completion
78%
Avg. time
6m 42s
Drop-off
Q7
01

Snapshot report

Top findings ranked by impact, with verbatim quotes. Completion rates, drop-off points, structured respondent profiles, scoring across categories you defined upfront. Generated automatically.

02

Deep Analytics Chat

An AI analyst that has read every transcript. Ask in plain English: "show me all respondents with high purchase intent", "what blocked enterprise leads", "draft a section for my investor deck on churn."

03

Raw transcripts

Exportable as JSON or CSV. Plug into Cloud Code, n8n, or your own pipeline. We give you the data, you decide what to do with it.

Every insight in every Snapshot links back to the verbatim quote that produced it. No black box.

Why a form can't do this

The agent isn't running a script.
It's running on three things at once.

Your business goal.

Loaded once at setup. The agent knows what decision you're trying to make — and steers the conversation toward it.

Your domain context.

Loaded once. For a B2B SaaS company, that's how your buyers talk about onboarding, procurement, renewals, integrations. For a consumer app, different vocabulary entirely. You don't have to re-explain your industry on every interview.

The user's previous answers.

Read in real time. The agent skips questions that are already answered, goes deeper on the ones that matter, rephrases when something didn't land. No two conversations are the same.

That's why no form, even one with AI follow-ups, does this.

What you can run with us

Six research sprints we run today.

Churn discovery

Why are customers cancelling in their first 60 days?

Audience
Your churned and inactive users from the last 90 days.
Output
Top 3 fixable churn drivers, ranked, with quotes.
For
Product, Customer Success, Retention.
Run a churn sprint
PMF / beachhead

Which segment has the sharpest pain we solve?

Audience
Hypothesis-fit prospects from your CRM or community.
Output
Segment-by-segment evidence of pain, urgency, budget, and weak alternatives.
For
Founders, early-stage PMs.
Run a PMF sprint
Post-launch validation

Did the new feature actually land?

Audience
Users in the first 60 days of the rollout.
Output
What got used, what got skipped, what to fix in the next sprint.
For
Product, Growth.
Run a validation sprint
Onboarding friction

Where is time-to-value slowing down?

Audience
Customers in their first 30 / 60 / 90 days.
Output
Product friction vs organizational friction, mapped.
For
Onboarding, CS, Product.
Run an onboarding sprint
AI adoption & trust

What do users actually trust our AI features to do?

Audience
Users of your AI features (and the ones who turned them off).
Output
Trust segments, adoption blockers, governance signals required.
For
Product, AI team leads.
Run an AI trust sprint
Employee insights

Anything you'd ask in a 1:1 — at scale.

Audience
Your own team.
Output
Themes, sentiment, structured profiles.
For
HR, People Ops, founders.
Run an internal sprint

Don't see your question here? We've also done pricing research, message testing, partner discovery, and feature prioritization. Tell us what you're trying to decide.

Under the hood

Small details that change the data.

The difference between a survey people abandon and a conversation people finish lives in the details. Here's what we obsess over.

Automatic follow-ups

Respondents who drop off mid-conversation get gentle nudges to pick up right where they left off. Completion rates go up without anyone on your team chasing.

Any language, auto-detected

The agent speaks the respondent's language — auto-detected from their first reply, or set manually per audience. One study, many markets, no translation step.

Empathy & interview psychology

Built-in interviewing technique: open-ended phrasing, active listening cues, non-leading follow-ups. People actually talk — the way they would to a thoughtful human.

Goal-aligned closings

At the end of each conversation, the agent summarizes the top 2–3 themes from that respondent and offers a relevant next step — a consultation, a resource, or a follow-up call.

Honest fit check

A good fit for some teams. Not the right tool for others.

Good fit if

  • You have an existing audience: users, churned customers, employees, beta list, newsletter, CRM.
  • You're making a product, GTM, pricing, or messaging decision in the next 90 days.
  • You've wanted to do 30 user interviews and never had the calendar for it.
  • Your team is already working with AI tools and wants raw data they can analyze themselves.
  • Your audience is reachable by email, in-app message, or WhatsApp.

Not us if

  • You need cold panel recruitment. We don't supply respondents.
  • You need real-time customer support automation. We ask questions, we don't answer them.
  • You only need quantitative NPS dashboards.
  • You're looking for a smarter Typeform — we're a different category.
  • Your research has to follow a strictly standardized survey methodology.

Trust by default

The data is yours. The methodology is auditable.

Your audience, your data.

You bring the respondents. We don't sell or share any data. Transcripts and analysis stay in your workspace.

Every insight is clickable.

Each conclusion in the Snapshot links back to the verbatim quote that produced it. No black box. No 'the AI said so'.

You review before anything ships.

The interview flow, the questions, the scoring rubric — all reviewed and approved by you before a single user sees a message.

Human-in-the-loop where it matters.

The agent is constrained by the goal and rubric you set. It doesn't freestyle. It doesn't lead respondents. It doesn't generalize beyond what people actually said.

GDPR-friendly setup. SOC 2 in progress. Custom data retention and PII redaction available on request.

How to start

Start with one decision. We'll handle the rest.

Beta offer
First project: $0.

In exchange for structured feedback on the platform. We're in active beta and learning as fast as you are.

Start a research sprint
Reach out on LinkedIn
We typically reply within a day.
How a first project looks
  1. 01
    One use case. One audience. One decision.

    Pick the sharpest open question your team is asking. We'd rather do one thing well than five things vaguely.

  2. 02
    Our AI agent drafts the interview. You review.

    Based on your goal (and any context you upload), the agent designs the opening, key questions, follow-up logic, and scoring rubric. You sign off before anything ships — no scripts or branches for you to wire up.

  3. 03
    We run with a small respondent group.

    Limited audience, web link or messenger only, no integrations needed for a first run.

  4. 04
    Raw data immediately. Insight in minutes.

    Transcripts stream in as people finish. Analytics on-demand, or auto-generated after a threshold of completed interviews. Real value the moment respondents hit send — no waiting on a weekly report.

  5. 05
    Decide together what's next.

    If it worked, we discuss what a longer engagement looks like. If it didn't, you tell us and we learn.

FAQ

Questions we hear.

No, and it's the question we get most often.

ChatGPT is great at conversation but bad at research. It will follow whatever the respondent says, lose track of the goal, ask leading questions, and forget what's been covered.

HaveTheWhy is constrained by three things: the business goal you load, the scoring rubric you approve, and a coverage map that makes sure every respondent is asked the questions that matter. The agent adapts inside that frame — it doesn't freestyle outside it.

One decision, one audience, $0.

Tell us what you're trying to learn. We'll get back to you in a day.

Start a research sprint