PROPHET is a solution of Human-in-the-loop (HITL), a branch of artificial intelligence, that takes advantage of human and machine intelligence to create machine learning models progressively.

In this way, it is possible to implement an immediate solution, which uses people to train the algorithm in a progressive and intuitive way from 0 to the desired quality levels.

PROPHET uses IBM DSX Local.

How does it work?

1. First, humans label the input data. This provides the model with large amounts of training data. An automatic learning algorithm learns to make decisions from these data.

2. Then, humans refine the model. This can happen in different ways, but commonly, humans score the quality of the output to avoid overfitting, to teach a classifier about limit cases, or about new classification categories.

3. Finally, people can test and validate a model by scoring their results, especially in places where an algorithm has no certainty or has too much confidence in an incorrect decision.

Solutions

Cognitive Solution for Call Center

KONECTA allows humans to intervene in a conversation a bot is having with a customer. The bot is automatically stopped if any problem came up, after that an assistant take the conversation control and turn on the bot whenever he wants. BOT > HUMAN > BOT

Conversational knowledge database

Ask questions just like you would ask a person.
The user ask something and the chatbot looks for the answer in the knowledge database.

If the information for answering the question is not available in the base the chatbot will trigger a process.

2 billion people use WhatsApp every day. Customers want a direct business communication channel throgh the apps they already use.

Is your company in WhatsApp?