NEURALOOT is a suite of Max for Live devices built around the neural oscillator, which is a wavetable oscillator whose shape is generated by a neural network (the decoder section of an autoencoder, if you want to get more technical).
This in house neural network was trained on thousands of waveforms, and it learnt how to encode in the most exact way any shape in just eight parameters, that make up the latent space of the network. Each combination of these parameters is able to reconstruct a waveform, and changing gradually one or more components of the latent space lets you seamlessly morph a shape into another one.
The complexity and the properties of the generated shapes makes them perfect to be used as modulators, sequencers and drone generators. NEURALOOT takes advantage of the neural oscillator to create all this nice stuff.