Customer discovery, with receipts

Find the problem worth building for.

Turn messy interview notes into ranked problem evidence. DeepWell keeps every conclusion attached to what a customer actually said.

01 / Frame the search

Bring the conversation. Name your hunches.

Interview statements10/24 lines

Maya, operations manager: Every Monday we count each coffee pack on three shelves and copy the totals into a spreadsheet.
Maya: Last month the count took two staff almost three hours because product names on the shelf did not match the spreadsheet.
Maya: When numbers look wrong, I take shelf photos and ask each branch manager to confirm stock in WhatsApp.
Maya: We ran out of our best-selling Gayo packs twice last month and estimate we lost about Rp 4 million in sales.
Maya: I usually guess the next purchase from last week's sales because the count is already old by the time I receive it.
Maya: A dashboard with many charts would look nice, but I would still check the physical shelf before ordering.
Rafi, branch lead: Yesterday I called the warehouse after a delivery was two hours late because the tracker only said "on the way."
Rafi: Late delivery is frustrating, but I can normally borrow two boxes from the nearby branch.
Rafi: What hurts most is discovering at opening time that yesterday's spreadsheet said we had stock when the shelf was empty.
Rafi: I spend about forty minutes each evening writing down low-stock items for the manager.

One statement per line. Speaker labels are welcome. Nothing is stored.

Suspected problem themes3/5 hypotheses

You supply the possible wells. The model tests the evidence against them.

Each statement is judged for

Problem theme
Evidence type
Concrete behavior
Urgency
Commitment

24 statement cap · 1 bounded request · 0 saved records