A research lab giving machines a sense of smell.
Machines can already see, hear, move, and speak. They still mostly cannot smell — and smell is chemical information the physical world is full of.
What we build
Aeralyte is building the full smell intelligence stack, not a single gas sensor: controlled air sampling, compact electronic-nose arrays, on-device AI, calibration protocols, and rigorous experimental methods. The intelligence lives in the loop between air, sensor, model, protocol, and deployment.
The ladder
Where we are now
We have designed the air-sampling protocol, the electronic-nose sensor array, the ESP32-S3 firmware, and the truth-gated analysis pipeline that turns sensor responses into smell fingerprints. The bench is live: the board is flashed and three sensor families are returning real readings. We are completing the rig and moving into controlled chamber experiments.
What we are chasing first: repeatable smell fingerprints on our own bench hardware — and the discipline to publish only what each stage of evidence supports.