Lick the Toad is an online web application that was designed to enable the collaborative participation of the audience during a live coding performance. The system allows remote interaction using WebSocket programming and real-time Open Sound Control communication between users' web browsers and coder's personal choice of performance environment (e.g., SuperCollider, Pure Data, MaxMSP). The system combines a user interface to train a machine learning system and generates prediction data using a regression algorithm apropos to the users' input. Once training is done, the application switches to the prediction mode and offers the visualization of OSC data, sonic events frequency information controlled by a Markov chain module embedded in the system. This allows one a hybrid exploration of the performance, affording one to turn the device into a fully controllable autonomous instrument while at the same time emitting data to any OSC client system in the network. In addition to the sound generation, other higher musical characteristics are modulated by the participants. That is, the tempo of the sequential repetition of the sonic events is modulated by the number of participants. This enables establishing an ongoing dialogue between audiences and performers and influences the decisions of the coding process, affording the audience to be at the forefront of the performance instead of observing the live coding process passively while reciprocally influencing one another.