March 6, 2026
Sniff test or whiff of hype?
Show HN: A trainable, modular electronic nose for industrial use
Hacker News sniffs a ‘digital nose’—half say game‑changer, half say it stinks of demo‑ware
TLDR: Sniphi launched a “digital nose” that reads air chemicals and uses AI to identify smells for industry, promising safer factories and fresher food. The crowd is split: enthusiasts push practical pass/fail uses with real savings, while skeptics doubt reliability outside lab conditions and crack jokes about smell‑to‑image AI.
Machines learned to see and speak—now a startup called Sniphi wants them to smell. Their “digital nose” uses gas sensors to read VOCs (tiny chemicals in the air) and an AI “brain” in Microsoft’s cloud to recognize scent patterns in real time. Cool demo? The comments fired up faster than a hot pepper sniff test.
A Sniphi team member (limel) went full business-mode, saying the tech is ready but the real win is “the first scalable use case” that proves ROI—aka money saved, not just noses wowed. The skeptics showed up early: one commenter says these devices often use off‑the‑shelf VOC sensors and “work in sterile conditions” but flop in the wild. Translation: great in a lab, chaos at a factory. Meanwhile, a pragmatic crowd rallied around a “find a wedge” play: pass/fail decisions where people already use smell—think catching bad ferments, leaky seals, or early mold—especially if the nose can beat humans and doesn’t need constant recalibration.
There was nostalgia too: one user built a DIY e‑nose a decade ago and hinted at medical angles. And the meme squad? “Can’t wait to run smell‑to‑image models,” joked another, imagining AI turning odors into pictures. If Sniphi’s platform really survives messy, real‑world air and proves savings in food, safety, or medicine, HN thinks it could be a big whiff—er, win. Until then, it’s a sniff-off between dreamers and doubters.
Key Points
- •Sniphi, founded by Antdata, introduces a modular Digital Nose platform for scent and gas recognition.
- •Sensors capture gas and VOC patterns, which ML models analyze to identify digital fingerprints in real time.
- •The system leverages Microsoft Azure services: IoT Hub, Azure Databricks, Power Apps, Power BI, and Azure SQL.
- •Models can be deployed on-device (edge) or in the cloud via APIs, enabling flexible implementations.
- •Target applications include food and beverage quality, safety and security, fragrance and cosmetics, and medical use cases.