The accelerating digitalisation of the economy has led to a situation where companies are increasingly embracing the power of artificial intelligence. According to a survey by the Fraunhofer Institute conducted in spring 2020, 75 per cent of the companies surveyed are planning to implement AI solutions, while 16 per cent are already using them to increase efficiency and reduce costs.
Innovations and services that analyse and improve corporate processes account for the largest share of Berlin's AI sector (36 per cent). They include a range of information and language processing systems that automate business processes, making communication with customers more efficient. With the help of AI systems, market data can be analysed to ensure marketing activities are implemented in a more targeted manner and the new methods of image and text recognition offer innovative ways to protect user data.
Customer excellence through digital language systems
In the digital space, customer demands have grown by leaps and bounds in recent years. For many users, the decisive factor in choosing a service is customer experience: companies are expected to provide 24/7 contactability, because efficient and transparent communication not only ensure satisfaction, but also trust. This places an increasing burden on customer service teams who not only have to cope with an enormous volume of customer enquiries, but also have to record each encounter as well as fulfil other administrative tasks. The time pressure involved all too often comes at the expense of quality.
To ease this burden, a number of Berlin-based companies, such as Rasa, Parlamind, Future of Voice, Solvemate and DialogShift, have developed various AI solutions with automatic speech recognition systems that can handle routine tasks such as call screening, chat messages and emails, or even take over an entire digital communication or customer service support system.
But how many times have you encountered an automated assistant that creates problems rather than solving them? Dr Tanja Klüwer from the Berlin start-up Parlamind believes that to achieve holistic customer communication you need a holistic solution. The company automates online communication through Task Automation and Assistive Support in the three main channels of customer service: email, telephone and chat. The system automatically evaluates incoming messages with total ease, based on the latest results from the built-in machine learning technology.
Another player in the conversational AI market in Berlin is Rasa. Pursuing the same goal but with a different approach, the company offers open source machine learning tools for developers and product teams to automate contextual multi-turn conversions, putting an end to chatbot frustration. Its AI solution offers one major advantage: The Rasa code is open and system-independent, allowing the software to be integrated into any corporate in-house infrastructure, which in turn ensures you can keep control of your data. The Rasa AI toolkit is now being used by thousands of developers and is also installed in some Fortune 500 companies.
Language assistants could hardly be more topical for organisations in the current global coronavirus pandemic. An example of this is a joint venture between the Munich-based PRIMO MEDICO and the founders of DialogShift in Berlin, who have developed a chatbot for Covid-19 crisis communication and provided it to the state-owned Vivantes Hospital Group free of charge. Within the first few weeks of the pandemic, the chatbot was answering over 1,000 questions per day in several languages. The idea behind the project was not to design a chatbot that could answer every single question about coronavirus, but instead to provide automated answers to frequently asked questions on the hotline. This relieves the customer service team of some of the work burden, without reducing response times - even in times of crisis.
© Uhura Digital GmbH for Berlin Partner for Business and Technology GmbH
Managing the data deluge
It is well known that data is key to digital success. But more data does not automatically mean more efficiency. Instead, many companies are faced with an overwhelming flood of internal and external data that can no longer be effectively processed using conventional methods. Here fast and efficient processes are required, which can be achieved through intelligent structuring and automated classification of documents and data sets.
The Berlin-based company dida specialises in process optimisation through AI-supported data analysis. The interdisciplinary team develops machine vision algorithms that can recognise structures in images and text that cannot be detected by the human eye. Their field of application ranges from insurance and e-commerce to real estate, health and weather forecasts. Machine learning technology can also make light work of some of the most tedious bookkeeping tasks. The company CANDIS, for example, provides automated solutions for invoice approval using AI. Thanks to their software, companies not only save time, but also minimise errors.
The ability to analyse unstructured data, allows large amounts of data to be evaluated quickly and automatically in order to provide data-based support for experts contemplating difficult decisions. The company SiaSearch, a start-up that emerged from the Berlin-based AI incubator Merantix, offers a comprehensive management platform for unstructured sensor data. These are automatically indexed and structured based on intuitive semantic attributes and keywords so that self-driving cars, for example, can make smart decisions in an emergency.
This predictive analysis is also used to implement the marketing activities of a company more precisely by analysing specific user behaviour and adapting products and services to these needs. However, internal data alone is often insufficient for making intelligent predictions using deep learning algorithms. Ambitious start-ups in the German capital like Datarade are tapping into this growing hunger for external, company, personal or machine-related data. The team sees its intelligence platform as an anchor in an extremely opaque data market that provides high-quality, GDPR-compliant data between buyers and sellers. Instead of wasting months looking for external data, it enables employees to efficiently identify suitable data providers and centralise sourcing.
Data usage meets data protection
High data quality is essential for innovations in the field of artificial intelligence - from self-driving cars to automated production and radiology imaging processes to combating online fraud and sophisticated marketing campaigns. However, the use of data should always be carried out hand in hand with the current data protection regulations. In Berlin, there are a growing number of companies engaged in data protection through AI.
The Berlin-based Statice offers a data anonymisation tool that enables companies to exploit the potential of their data while complying with the privacy regulations that protect personal data. Its solution, which is primarily used in the financial, health and insurance sectors, transforms this incoming data into smart synthetic data that is completely anonymous, but still preserves its original statistical properties. Another brand new innovation from the German capital is the Xayn search engine launched by the start-up Xayn. The company aims to position itself as an alternative to Alphabet's Google by offering a combination of data protection and convenience: user data is stored centrally on a smartphone, while an AI recognises user interests through elimination processes and tailors content specifically to the individual user.
The Deep Neuron Lab, on the other hand, focuses entirely on personal data stored by a company internally. Using state-of-the-art object recognition technology called Convolutional Neural Networks, the Anonymiser AU automatically identifies and anonymises any form fields that contain sensitive data and makes these documents legally usable. Under the current data protection regulations stipulated by GDPR, CCPA, APPI and CSL, visual data such as photo or video material are also recognised as particularly worthy of protection - with serious consequences for breach of privacy. However, engineers and developers in fields such as autonomous driving rely on image data from public areas to create the technology that enables vehicles to perceive their surroundings and recognise potential dangers early on. Brighter AI, a leader in visual data anonymisation, uses a deep learning algorithm that redacts personally identifiable information such as faces or license plates and replaces it with replica data generated by the system. This ensures that identities in public are protected and at the same provides data protection-compliant data sets for software development, marketing analyses and algorithm training, proving that data-intensive usage and data protection do not have to be mutually exclusive.
Increasing efficiency while lowering costs are the maxims of the modern, digitized business world, so the huge need for optimization and further development of internal processes is the logical consequence. With strong, diversified, AI-based innovations and services, however, the Berlin AI ecosystem is well on the way to having a say in the global business intelligence market.