About Usense
Better panels. Cleaner data. At a fraction of the cost.
Usense combines an AI panel leader that trains each panelist individually with a data collection layer that produces standardised, molecule-anchored sensory scores — without imposing the vocabulary that typically distorts them.
The challenge
No two noses are wired the same way.
Sensory perception is shaped by a lifetime of personal experience — the aromas of childhood kitchens, regional cuisines, cultural habits. These experiences create a deeply individual inner map of smells that no two people share exactly. Traditional panel training acknowledges this with a uniform curriculum: everyone learns the same references, in the same order, in the same room. It produces panels that can agree on a label — but not panels where each member has genuinely internalised why that label is correct for them.
Months of investment. Data that still varies.
Assembling a trained sensory panel is a significant undertaking — recruiting the right profiles, running weeks of screening sessions, booking facilities, coordinating schedules around a specialist panel leader. Once trained, panels require regular maintenance sessions to stay calibrated, and results still vary meaningfully between sessions and sites. For the time and cost involved, the reproducibility of the data rarely matches the ambition of the process.
Personalised training
An AI panel leader that knows each panelist by name.
Molecular references are not new to sensory science — professional panels have always been trained on them. What has never existed, until now, is training that meets each panelist where they are. We all perceive aromas through the lens of our own history: the food we grew up with, the places we have lived, the smells that marked us. Two people can detect the same molecule and reach for entirely different words to describe it — and both can be right. Usense maps those personal associations onto the universal reference, building a bridge between individual perception and shared scientific language. The AI learns each panelist's sensory profile — their strengths, their blind spots, their sensitivity thresholds — and tailors every session to close the gaps that matter most to them specifically.
Our AI coach guides each panelist through this reference set individually — adapting to their pace, reinforcing the aromas they struggle with, and refusing to move on until mastery is demonstrated. The result is a panelist who does not just know what vanilla smells like in theory. They know it precisely, reliably, and in their own sensory memory.
Standardised data collection
Scores you can compare. Words you can trust.
When panelists evaluate a product, Usense captures their responses and maps them back to the molecular reference framework — regardless of the words they use. A panelist who says "it smells like my grandmother's garden" and one who says "floral, slightly powdery" can be describing the same compound. Because every response is anchored to a molecule rather than a prescribed term, the data is inherently standardised across individuals, sessions, and languages.
This matters because traditional sensory evaluation carries a well-documented risk: when panelists are given a fixed descriptor list, the list itself shapes what they perceive. They select the closest available word rather than the most accurate one. Usense removes that constraint entirely. Panelists describe what they detect in their own words — and the platform handles the translation into structured, comparable data. The perception stays uncontaminated. The output stays consistent.
Molecule-anchored scores
Every evaluation maps to a universal molecular reference — making results comparable across panelists, sessions, sites, and time, without any manual harmonisation.
Free-form expression, structured output
Panelists describe what they detect in their own words. The AI maps their language to the reference framework — removing the distortion introduced by fixed vocabulary lists.
What changes
Panels that genuinely agree.
Calibrated individually against the same molecular ground truth, panelists reach consistent conclusions for the right reasons — not because they were told which words to use.
Data you can act on.
Molecule-anchored scores are reproducible across sessions, geographies, and time. Product decisions, reformulations, and quality benchmarks are finally grounded in something stable.
Training at a fraction of the cost.
The AI coach works one-on-one, at each panelist's own pace, anywhere. No specialist facilitator, no centralised sessions, no scheduling overhead — just rigorous, personalised calibration.
"The most honest sensory data comes from panelists who are free to say exactly what they perceive — and a system precise enough to make sense of it."
Usense · Augmented Sensory