Model Deployment

I am trying to work on deploying my first model to Heroku.

So I was referring some videos for it. I saw that some pre-processing like removing stopwords, lemmatizing etc are done while creating the initial model. After that the model is dumped using pickle.

However, when taking new input data there is no pre-processing of that data and directly the input is given to the model for prediction,

Can anyone help on this, not sure how to approach this.