Staying useful when the real world changes.

Sensor drift is one of the biggest barriers in electronic noses. Aeralyte's long-term moat is calibration data: every controlled experiment and field deployment teaches the system how smell changes across environments.

Technical terms: drift compensation, calibration transfer, humidity correction, confounder-invariant learning.

More experiments produce more calibration data; better drift correction produces more reliable fingerprints; more reliable fingerprints make deployments more valuable — which produces more experiments.

Humidity and temperature (SHT40) are tracked as first-class confounders, not afterthoughts, so the model can learn what to ignore.

Two checks speak to robustness. A sensor-family ablation showed the baseline holds up even with a family removed — no single sensor is a crutch. An adversarial label-shuffle collapsed accuracy to chance, confirming the model learns real structure, not a leak.

This is the question we care about most: real drift across humidity, sensor aging, and background air — the central target of the bench and chamber experiments.

• 01AERALYTE_RESEARCH_LAB_BRIEF.md — pillar 5, drift & reality; the data flywheel.
• 02PHASE1_CONTROLLED_ENOSE_EXPERIMENT.md — sensor-family ablation & label-shuffle checks.
• 03Kimi_Agent · Sensor Evolution & Power Trade-offs — drift & calibration.
• 04GitHub — XoAnonXo/aeralyte.