The new normal that changes how we innovate with AI

In the industry, there is a burgeoning need for intelligent systems working out-of-the-box. This is especially true for the young AI startups aiming to disrupt, needing at the same time a solid ground to start from and the flexibility to adapt.

These two contradicting requirements are now fulfilled, due to the freely available and massive pre-trained models like BERT.

This paradigm shift allows to rapidly develop and fine-tune sophisticated AI models based on the business needs. For example, we can use BERT to do the following:

Text Classification

Text classification can be used to automate the following tasks:

• Sentiment Analysis: beneficial to understand our customers;
• Intent classification: automatically route a customer request to the right agents;
• Customer categorization: group social media users into cohorts.

Named-Entity Recognition (NER)

Probably one of the most useful tasks for business intelligence:

• Knowledge extraction: understand the written text and link it against databases to show results to the user;
• Chatbots: understand customers better.

Text Summarization

Automatic text summarization can be applied for:

• Marketing and SEO: summarize the content of the competitors to understand it;
• Financial reports: extract meaningful information for better decisions
• Legal contracts: understand them in simple words

If you are interested to know more details, please refer to my blog post