Company

OdorPi - Our First Steps

Sep 1, 2025

min read

For the past six months, OdorPi has been operating in stealth—designing, testing, and obsessing over a question that could change how we interact with technology: What if computers could smell?

Over the next few months, we’re lifting the curtain a little to share what we’ve been up to, what we’re building—and why it matters.

Some Background Context

Since modern-day computers were invented, we’ve digitized 3 of our 5 senses - sight, hearing and touch.

With the camera and microphone, computers can now see and hear. Because of that, Skype was possible - and now, we can see and hear our loved ones thousands of miles away at the click of a button, 24/7.

With the touchscreen, computers learned to feel. Suddenly, glass was no longer just a display—it became a canvas for creativity and a bridge for interaction. We could pinch to zoom, swipe to share, and tap to connect.

Sensors give computers the ability to understand the physical world. When we give computers this ability, the applications are extremely valuable. Computers can see inside our body (MRI, X-Rays). They can detect deadly Carbon Monoxide or Methane gas leaks. They can help us navigate in a new city. They help pilots fly safer. They can detect a car crash and trigger an airbag… and much more.

Smell, i.e. Olfaction has not been digitized. We at OdorPi believe there’s a huge, untapped opportunity here to better our lives by giving computers the ability to smell.

Our “Why” - Better Healthcare.

Long before visible symptoms of disease appear, our bodies warn us. Subtle changes in our sweat, breath or urine point to changes in the underlying metabolic state of our body. We’ve known this for a long time. Hippocrates, the ancient Greek physician, utilized his sense of smell to diagnose diabetes.

More recently, trained dogs have been shown to detect many forms of cancer simply by sniffing bodily fluid samples. We did a quick meta-analysis on academic literature and the results highlight the olfactory prowess of our canine friends. In many double-blind studies, they match or beat standard-of-care diagnostic tools (i.e. MRI, X-Ray, physical examination, etc.).

For the most common types of cancers, the results are presented below:

It’s worth noting that the dogs are making quick judgements (under 10s per sample) as to whether a sample is positive for a certain cancer indication. Factoring out wait queues, our best tools take hours from test to result.

Perhaps due to our relatively weak sense of smell and the complexity involved in mapping the subjective world of scent, the opportunity to digitise smell has not been adequately explored. While there are leading scientists researching smell, the collective investment into the space is a fraction of what it should be.

What we’re doing

We’re developing an artificial electronic nose (E-nose) and pairing it with our in-house AI model to interpret and recognize smells. It’s not a single-purpose gas sensor aimed at detecting one gas, nor is it a mass spectroscopy device e.g. GC-MS, aimed at detecting individual molecules.

What’s Next

We’re heads down into R&D now – developing our in-house sensitive membranes and scaling up our AI model training pipeline.

Company