of macroscopic systems, reductionism is always employed: molecules are the smallest compositive units, and a whole (system) is nothing but the sum of its individual parts (molecules). [1] And based on this approach, many scientific rules and theorems such as the Beer-Lambert law are developed through experimental data in dilute solutions to avoid the interference of intermolecular interactions. [2,3] However, the behaviors and properties of molecules will be largely changed in concentrated solutions or in their aggregate state. For example, many aromatic compounds emit intensely in dilute solutions but become weakly emissive in the aggregate state, such a phenomenon is referred to as aggregation-caused quenching. [4] On the other hand, some molecules would act differently and show enhanced light emission when aggregated, such a behavior is called aggregation-induced emission (AIE). [5] These aggregates with mesoscale are important mediators for the transfer and amplification of macroscopic properties. Therefore, the paradigm shift from reductionism to aggregation science helps the study of exploiting and regulating their new properties. [6] Molecular aggregates with environmental responsive properties are desired for their wide practical applications such as bioprobes. Here, a series of smart near-infrared (NIR) luminogens for hyperlipidemia (HLP) diagnosis is reported. The aggregates of these molecules exhibit a twisted intramolecular charge-transfer effect in aqueous media, but aggregation-induced emission in highly viscous media due to the restriction of the intramolecular motion. These aggregates, which can autonomously respond to different environments via switching the aggregation state without changing their chemical structures are described, as "smart aggregates". Intriguingly, these luminogens demonstrate NIR-II and NIR-III luminescence with ultralarge Stokes shifts (>950 nm). Both in vitro detection and in vivo imaging of HLP can be realized in a mouse model. Linear relationships exist between the emission intensity and multiple pathological parameters in blood samples of HLP patients. Thus, the design of smart aggregate facilitates rapid and accurate detection of HLP and provides a promising attempt in aggregate science.