AI Startup Claims Human-Level Smell Intelligence

A New Jersey fragrance startup says its model has matched trained human perfumers in judging scents.

Updated on Nov. 22, 2026 in Artificial Intelligence

AI Startup Claims Human-Level Smell Intelligence

Live Poll

Do you trust AI-designed products to work as well as human-made ones?

A New Jersey AI fragrance startup claims its model reached human-level olfactory intelligence, outperforming in-house perfumers at an unnamed consumer goods company in October 2026. The company's founder, Alex Wiltschko, spun the work out of his Google Brain research in 2023.

Why it matters

If confirmed, the approach could expand how machines detect and describe odors, an area with few existing digital datasets. Wiltschko has suggested future uses could include airport threat detection and disease diagnosis.

The dataset pairs about 30 experienced human sniffers and roughly 70 trained labelers with spectrometer and chemical sensor analysis run around the clock. The company says the system identified eight molecules that repel disease-carrying insects better than DEET, a claim not yet independently confirmed.

The players

Alex Wiltschko

Founder of the New Jersey AI fragrance startup, formerly a researcher at Google Brain.

Gates Foundation

Philanthropic organization founded by Bill and Melinda Gates that funded the company's early work and later insect research.

Procter & Gamble

Multinational consumer goods company for which the startup has created retail fragrances.

The details

The approach cross-references human descriptions of smells, made using a standardized fragrance vocabulary, with chemical analysis of fruit, flowers and pantry staples. The company has created retail fragrances for Procter & Gamble and received $5 million at launch from the Gates Foundation, plus a later $3.5 million grant.

Timeline

  1. 2023: Work spun out of Google Brain research.

  2. October 2026: Model reportedly outperformed in-house perfumers at a consumer goods company.

  3. November 18, 2026: MIT Technology Review published an interview with Wiltschko.

The Tech Race

The claim places an AI startup alongside established repellent benchmarks such as DEET, which has long served as the standard for insect protection. It also extends the broader push to use machine learning on sensory data that has historically lacked large digital datasets.

If the technology reaches commercial use, consumers could see new fragrances and insect repellents developed through AI analysis of smell. Practical adoption, pricing and independent verification of the company's claims remain unclear.

The takeaway

Machine learning is beginning to extend beyond sight and sound into smell, a sense with few digital datasets. Readers should watch for independent verification before accepting the claims.

Further reading

Explore more Artificial Intelligence coverage for context on machine learning advances.

Source note: This article includes information reported by Teknoloji Haberleri - ShiftDelete.Net.

Live Poll

Do you trust AI-designed products to work as well as human-made ones?