CASE STUDIES

CHATBOTS FOR BUSINESS PROCESS AUTOMATION

Use Case: Customer Service Chatbot

Context

Easy and direct customer support through a chat interface on the website is becoming a major trend. Customer services agents can engage more meaningfully with customers and address issues efficiently.

Challenge

Many customers ask the same common questions and service agents keep repeating the same answers.

Solution

We created an automated chatbot that is able to respond to 80% of commonly asked questions 24/7. The chatbot reacts flexible to customer requests with human-like contextual conversations. A handoff to customer service agents is possible when necessary.

IMAGE ANALYSES SYSTEMS

Use Case: Detecting Cell Components

Context

Microscopes in medical research acquire a lot of images that hold valuable scientific information in an unstructured format. For meaningful biological analyses, the objects in these images need to be fully labelled.

Challenge

Labelling cell components manually is too time-consuming.

Solution

We created an automated mitochondria classifier using deep learning technology. We applied state-of-the-art research to detect cell components automatically with minimal required human input. Although, the classifier has been trained on a single dataset, it also adapts to other datasets.

WORKSHOPS ON UNDERSTANDING AI

Use Case: Building a Self-Driving Car

Context

Autonomous driving is the next big challenge for the automotive industry. What seemed infeasible a couple of years ago is becoming reality quickly. But what fuels this innovation?

Challenge

Understanding the basic AI principles behind self-driving cars and building a working prototype.

Solution

As part of our workshop we explained and implemented the basic AI principles behind self-driving cars using neural networks. A modified RC model car is equipped with a camera and computer and is trained to steer along a race track autonomously. The techniques conveyed during the workshop were based on research publications from Nvidia.