ALEX · AI USER RESEARCHER
Your users know why. Ask them in the moment.
AI interviews, inside your product
Alex asks the question you would ask in person, right when it matters: why they came, why they are leaving, what they were trying to do. Then it turns the answers into themes you can act on.
- paid subscription · card required
- One snippet · works with your stack
- Consent before every interview
- If eligible, cancel before day 14 and pay nothing · 30-day refund after subscribing
Northstar workspace
Product analytics
Activation rate
42.8%
Active users
18,420
Interviews
128
New signals
24
Alex
liveA quick product question
Can I ask you one quick question?
Alex only starts recording after the user says yes.
Three questions Alex asks for you
Start with the one that hurts most. Each is a short chat, not a form.
What brought you here?
Signed-out visitors on your homepage or pricing page
Alex asks what they came for, what was going on right before, and what they use today. You hear why people really arrive, in their own words.
You get: Themes, plus one clear reason per visitor.
Running on specky.space since 7 November 2026. It records the conversation and nothing else. Ask us to switch it on for your campaign.
Why are you leaving?
Your cancel page, or the link in your cancellation email
Alex asks what made them decide, what they will use instead, and what would have kept them. It listens and never pitches.
You get: The real reasons people leave, grouped into themes.
Send the interview link to people who cancelled, or paste the widget on your cancel page.
What were you trying to do?
New users, in their first sessions
Alex asks them to use your product as they normally would, then what they were trying to get done and where they got stuck. With their consent it also notes which parts they clicked: names only, never what is on screen.
You get: A transcript, an action log, the task outcome and themes.
We run this inside the Specky app, for accounts in their first 14 days.
We run Alex on our own site
Since 7 November 2026, a signed-out visitor to specky.space can get one question from Alex: what brought them here.
- What Alex says. An actual opening from our homepage (the wording varies): “Hi there, I'm Alex—thanks for stopping by Specky today. What brought you here, and what were you hoping to get done?”
- What happens to the answer. It is not only stored. It fills in the first question of our signup setup, “What brought you to Specky today?”, so nobody has to say it twice. They can change it.
- What we keep. The conversation, for 90 days, and nothing about what the visitor does on the site. Consent comes first.
- What we have learned so far. Nothing worth quoting yet. It went live on 7 November, and we will not show you numbers from a handful of chats. Once there are enough real conversations, we will publish what Alex heard and what we changed because of it, here.
You are asked for consent first. It is a real chat, and we read it.
How Alex works
From snippet to insights in three steps — no migration, no setup beyond a test.
1. Add one snippet
Paste a script tag before </body> and point it at a test campaign. Choose which feature to ask about and what % of visitors to sample.
2. Alex interviews live
A small widget asks a random user for consent, then Alex chats with them. In a study it also notes which parts of your product they click, by name; in a plain conversation it records only the chat.
3. Insights land in your workspace
Transcripts, themes and sentiment arrive in Specky as research nodes (plus the action log and task outcome for studies) — ready to feed your roadmap.
Alex vs. in-app testing alternatives
The difference: they show you what users did, or count what they ticked. Alex also asks them why — while they're doing it.
| Tool | Their approach | With Alex |
|---|---|---|
| Hotjar | Shows what users do. It never asks why. | Alex interviews users while they act — you get the why behind the click. |
| Maze | Tests a prototype you build in a separate tool. | Alex runs in your real product, on the real feature, with real users. |
| UserTesting | Scheduled sessions with recruited participants. | Alex interviews a random slice of your actual users, on demand, at 5% of the cost. |
| PostHog Surveys | Fixed questions: ratings, choices, open text. Strong for counting; a one-line answer is not followed up. | Alex follows up like a person would, then groups what it heard into themes. It works alongside PostHog. |
Consent-first by design
Alex never records anything before a user explicitly agrees, and every session stores a consent timestamp. Respondents never need an account, and you stay in control of the sample rate. Interviews land as private research in your workspace — GDPR-respecting by construction.
Frequently asked questions
What is Alex?
Alex is an AI interviewer that runs short conversations inside your own product. A small widget asks a user for consent, then Alex chats with them about why they came, why they are leaving, or what they are trying to do. The interview, themes and sentiment land in your Specky workspace.
Does Alex record clicks?
Only in a study, and only after the user agrees. A study notes which parts of your product they click, by name, never the text on screen or what they type. A plain conversation records the chat and nothing else. Studies and link interviews are available now; conversation mode is in early access.
Do I have to migrate from my current tools?
No. Alex works alongside Hotjar, Maze, UserTesting, PostHog, or anything else you already use. You just add one script tag and a test campaign — no moving data, no replacing your stack.
What about consent and privacy?
Nothing is recorded until the user explicitly agrees. Consent is required before the widget starts the interview, and every session stores a consent timestamp. No respondent ever needs to create an account.
What sample of users sees the widget?
You control the sample rate — by default Alex asks 5% of visitors. Each user is only asked once. You can target a specific feature or a specific percentage.
What do I get back?
A transcript, extracted themes and sentiment for every interview. Studies add an action log and the task outcome (completed, failed or abandoned). Link interviews also ask for a 0–10 recommend score. Everything is stored in your Specky workspace as research, so it feeds your roadmap.
Do you use Alex yourselves?
Yes. Since 7 November 2026 on specky.space (a one-question chat for signed-out visitors) and inside the Specky app (for accounts in their first 14 days). It is too early for results. We will publish what we learn once there are enough real conversations to be worth reading.
Ask your users, then build what the evidence supports.
Alex is live inside Specky. Create a test campaign, paste the snippet, and your first interview can be running today.