Luc.so
AI customer interviews, run with discipline.
The story
Overview
Lùc runs customer interviews the way a disciplined researcher would. It asks about what people actually did rather than what they say they would do, follows up on every answer in real time, and refuses to count a compliment as evidence. It's built for research, UX, and brand agencies that need to run more interviews per project without dropping the craft. You set up the study, Lùc conducts the conversations and holds each one to Mom Test and Jobs-to-be-Done discipline, and you get back clean material to synthesize. Currently in private beta.
Key features
Mom Test discipline
Asks about real past behavior and never plants the answer inside the question, so it doesn't lead the witness.
Refuses compliments
Treats "that sounds great" as noise and digs for what the person actually did.
Adaptive follow-ups
Reads each answer and asks the right next question instead of reading a fixed script
Jobs-to-be-Done probing
Finds the job a customer hired a product to do and the moment they switched.
Consistent structure
Section order, openings, and follow-up limits are set by the engine, so every interview holds the same standard.
Runs at volume
Conducts many interviews in parallel without a researcher in every call.
Use cases
- 1
Research agencies
Take on more interview volume per project without adding researcher hours.
- 2
UX research teams
Run discovery and usability interviews with the same question discipline every time.
- 3
Brand strategists
Get at the real reasons people switch, stay, or leave.
- 4
Founders and product teams
Test whether a problem is real with customers before building.
FAQ
How is this different from a survey?
A survey records fixed answers. Lùc holds a conversation: it follows up on what each person says, asks for the specific last time it happened, and won't take a vague or flattering reply as data.
Does it replace researchers?
No. It runs the interviews and keeps them on method, so your researchers spend their time on study design and synthesis instead of moderating every call.
What is the Mom Test discipline you mention?
A set of rules for not leading people: ask about real past behavior, keep the answer out of the question, and treat compliments as noise instead of signal.
Can I use it now?
Lùc is in private beta. You can join the waitlist on the site and we'll follow up about access.
Tech stack & tags
Links
Feedback & Discussion
Discussion
The founder is asking for feedback
Would you trust an AI to run customer interviews you'd normally moderate yourself? And does the discipline angle (no leading questions, compliments treated as noise) sound genuinely useful?
Sign in to leave a comment or rating.
No comments yet. Be the first.
More AI tools
A URL for everything your agents make. The Google Drive for agents.